mirror of
https://github.com/run-llama/llama_cloud_services.git
synced 2026-07-19 00:04:30 -04:00
Compare commits
38 Commits
| Author | SHA1 | Date | |
|---|---|---|---|
 Jerry Liu
|
ff038c9d02 | ||
|
Jerry Liu
|
bd6133fe9a | ||
|
Jerry Liu
|
1c0f72953e | ||
|
Jerry Liu
|
e58b40b34c | ||
|
Jerry Liu
|
0db05b9b96 | ||
|
Jerry Liu
|
a8a191ae87 | ||
|
Hemant Malik
|
4d92775aa8 | ||
|
Jerry Liu
|
477847111e | ||
|
Adam Reichert
|
efbcfb1d2e | ||
|
Logan
|
a9b01c761c | ||
|
Pierre-Loic Doulcet
|
dac2f7c84e | ||
|
Logan Markewich
|
478142e509 | ||
|
Logan Markewich
|
e76d5ba679 | ||
|
Huu Le
|
23dc9c0f68 | ||
|
Sourabh Desai
|
8b96176d8a | ||
|
Logan
|
1bbf5f4823 | ||
|
Pierre-Loic Doulcet
|
2f57682035 | ||
|
Jerry Liu
|
4c05160f98 | ||
|
Jerry Liu
|
3e9ad64a7f | ||
|
Jerry Liu
|
21fa19c73b | ||
|
Pierre-Loic Doulcet
|
58257d546b | ||
|
Adam Reichert
|
5e99d810fd | ||
|
Jerry Liu
|
a2dc717d85 | ||
|
Logan
|
87062e6ca8 | ||
|
Jerry Liu
|
ae3a21c5ff | ||
|
Ravi Theja
|
ccebb8a2fa | ||
|
Pierre-Loic Doulcet
|
2d21d6e688 | ||
|
Logan
|
270d96b7e3 | ||
|
Ajit Mistry
|
ea21daa96f | ||
|
dependabot[bot]
|
42845f8d07 | ||
|
dependabot[bot]
|
8b63ae9c46 | ||
|
Pierre-Loic Doulcet
|
173060dc50 | ||
|
Pierre-Loic Doulcet
|
d19b35cd48 | ||
|
Jerry Liu
|
0c83fbd679 | ||
|
Jerry Liu
|
6ae9c1d9cb | ||
|
Jerry Liu
|
27523b657a | ||
|
Jerry Liu
|
56d73c1a3f | ||
|
Jerry Liu
|
0d2ad9faab |
@@ -72,7 +72,7 @@ repos:
|
||||
args:
|
||||
[
|
||||
"--ignore-words-list",
|
||||
"astroid,gallary,momento,narl,ot,rouge,nin,gere,te,inh",
|
||||
"astroid,gallary,momento,narl,ot,rouge,nin,gere,te,inh,vor",
|
||||
]
|
||||
- repo: https://github.com/srstevenson/nb-clean
|
||||
rev: 3.1.0
|
||||
|
||||
@@ -6,6 +6,8 @@ LlamaParse directly integrates with [LlamaIndex](https://github.com/run-llama/ll
|
||||
|
||||
Free plan is up to 1000 pages a day. Paid plan is free 7k pages per week + 0.3c per additional page.
|
||||
|
||||
Read below for some quickstart information, or see the [full documentation](https://docs.cloud.llamaindex.ai/).
|
||||
|
||||
## Getting Started
|
||||
|
||||
First, login and get an api-key from [**https://cloud.llamaindex.ai ↗**](https://cloud.llamaindex.ai).
|
||||
@@ -53,6 +55,34 @@ documents = await parser.aload_data("./my_file.pdf")
|
||||
documents = await parser.aload_data(["./my_file1.pdf", "./my_file2.pdf"])
|
||||
```
|
||||
|
||||
## Using with file object
|
||||
|
||||
You can parse a file object directly:
|
||||
|
||||
```python
|
||||
import nest_asyncio
|
||||
|
||||
nest_asyncio.apply()
|
||||
|
||||
from llama_parse import LlamaParse
|
||||
|
||||
parser = LlamaParse(
|
||||
api_key="llx-...", # can also be set in your env as LLAMA_CLOUD_API_KEY
|
||||
result_type="markdown", # "markdown" and "text" are available
|
||||
num_workers=4, # if multiple files passed, split in `num_workers` API calls
|
||||
verbose=True,
|
||||
language="en", # Optionally you can define a language, default=en
|
||||
)
|
||||
|
||||
with open("./my_file1.pdf", "rb") as f:
|
||||
documents = parser.load_data(f)
|
||||
|
||||
# you can also pass file bytes directly
|
||||
with open("./my_file1.pdf", "rb") as f:
|
||||
file_bytes = f.read()
|
||||
documents = parser.load_data(file_bytes)
|
||||
```
|
||||
|
||||
## Using with `SimpleDirectoryReader`
|
||||
|
||||
You can also integrate the parser as the default PDF loader in `SimpleDirectoryReader`:
|
||||
@@ -87,6 +117,10 @@ Several end-to-end indexing examples can be found in the examples folder
|
||||
- [Advanced RAG Example](examples/demo_advanced.ipynb)
|
||||
- [Raw API Usage](examples/demo_api.ipynb)
|
||||
|
||||
## Documentation
|
||||
|
||||
[https://docs.cloud.llamaindex.ai/](https://docs.cloud.llamaindex.ai/)
|
||||
|
||||
## Terms of Service
|
||||
|
||||
See the [Terms of Service Here](./TOS.pdf).
|
||||
|
||||
Binary file not shown.
@@ -0,0 +1,529 @@
|
||||
{
|
||||
"cells": [
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "c148b65e-e8a6-476e-86ba-bf6a73d479c7",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"# RAG over the Caltrain Weekend Schedule \n",
|
||||
"\n",
|
||||
"<a href=\"https://colab.research.google.com/github/run-llama/llama_parse/blob/main/examples/caltrain/caltrain_text_mode.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
|
||||
"\n",
|
||||
"This example shows off LlamaParse parsing capabilities to build a functioning query pipeline over the Caltrain weekend schedule, a big timetable containing all trains northbound and southbound and their stops in various cities.\n",
|
||||
"\n",
|
||||
"Naive parsing solutions mess up in representing this tabular representation, leading to LLM hallucinations. In contrast, LlamaParse text-mode spatially lays out the table in a neat format, enabling more sophisticated LLMs like gpt-4-turbo to understand the spacing and reason over all the numbers.\n",
|
||||
"\n",
|
||||
"**NOTE**: LlamaParse markdown mode doesn't quite work yet - it's in development!"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "ef115dbe-b834-4639-828e-e2c11aef710b",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## Setup\n",
|
||||
"\n",
|
||||
"Download the data."
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "e6ae2e38-30c9-4865-aa13-47780bc3848f",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"import nest_asyncio\n",
|
||||
"\n",
|
||||
"nest_asyncio.apply()"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "335ce1d0-757a-4f09-846e-21c409768871",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"!wget \"https://www.caltrain.com/media/31602/download?inline?inline\" -O caltrain_schedule_weekend.pdf"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "45fa9120-65bb-4772-9db7-53e7cecf9adc",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## Initialize LlamaParse\n",
|
||||
"\n",
|
||||
"Initialize LlamaParse in `text` mode which will represent complex documents incl. text, tables, and figures as nicely formatted text."
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "54aa9579-84d4-49bc-ab54-5474e69c1188",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stderr",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"/Users/jerryliu/Programming/llama_parse/.venv/lib/python3.10/site-packages/tqdm/auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n",
|
||||
" from .autonotebook import tqdm as notebook_tqdm\n"
|
||||
]
|
||||
},
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"Started parsing the file under job_id 5f73353a-1f4b-480d-9eea-58d1d22b75f6\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"from llama_parse import LlamaParse\n",
|
||||
"\n",
|
||||
"docs = LlamaParse(result_type=\"text\").load_data(\"./caltrain_schedule_weekend.pdf\")"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "602756b2-9ea1-4519-a8e3-c773ec624205",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"Take a look at the below text (and zoom out from the browser to really get the effect!). You'll see that the entire table is nicely laid out."
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "4928281a-591a-4653-b451-b2b8112a7101",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"ZONE 2ZONE 3ZONE 4ZONE 4 ZONE 3ZONE 2ZONE 1ZONE 1\n",
|
||||
" Printer-Friendly Caltrain Schedule\n",
|
||||
" Northbound – WEEKEND SERVICE to SAN FRANCISCO 2XX Local\n",
|
||||
"\n",
|
||||
"\n",
|
||||
" Train No. 221 225 229 233 237 241 245 249 253 257 261 265 269 273 *277 *281\n",
|
||||
" Service Types L2 L2 L2 L2 L2 L2 L2 L2 L2 L2 L2 L2 L2 L2 L2 L2\n",
|
||||
" Tamien 7:12a 9:05a 10:05a 11:05a 1:05p 3:05p 5:05p 7:05p 9:05p 11:05p\n",
|
||||
" San Jose Diridon 7:19a 9:12a 10:12a 11:12a 12:12p 1:12p 2:12p 3:12p 4:12p 5:12p 6:12p 7:12p 8:12p 9:12p 10:19p 11:12p\n",
|
||||
" Santa Clara 7:25a 9:18a 10:18a 11:18a 12:18p 1:18p 2:18p 3:18p 4:18p 5:18p 6:18p 7:18p 8:18p 9:18p 10:25p 11:18p\n",
|
||||
" Lawrence 7:31a 9:24a 10:24a 11:24a 12:24p 1:24p 2:24p 3:24p 4:24p 5:24p 6:24p 7:24p 8:24p 9:24p 10:31p 11:24p\n",
|
||||
" Sunnyvale 7:35a 9:28a 10:28a 11:28a 12:28p 1:28p 2:28p 3:28p 4:28p 5:28p 6:28p 7:28p 8:28p 9:28p 10:35p 11:28p\n",
|
||||
" Mountain View 7:40a 9:34a 10:34a 11:34a 12:34p 1:34p 2:34p 3:34p 4:34p 5:34p 6:34p 7:34p 8:34p 9:34p 10:40p 11:34p\n",
|
||||
" San Antonio 7:43a 9:37a 10:37a 11:37a 12:37p 1:37p 2:37p 3:37p 4:37p 5:37p 6:37p 7:37p 8:37p 9:37p 10:44p 11:37p\n",
|
||||
" California Ave 7:48a 9:42a 10:42a 11:42a 12:42p 1:42p 2:42p 3:42p 4:42p 5:42p 6:42p 7:42p 8:42p 9:42p 10:48p 11:42p\n",
|
||||
" Palo Alto 7:52a 9:46a 10:46a 11:46a 12:46p 1:46p 2:46p 3:46p 4:46p 5:46p 6:46p 7:46p 8:46p 9:46p 10:53p 11:46p\n",
|
||||
" Menlo Park 7:55a 9:50a 10:50a 11:50a 12:50p 1:50p 2:50p 3:50p 4:50p 5:50p 6:50p 7:50p 8:50p 9:50p 10:56p 11:50p\n",
|
||||
" Redwood City 8:01a 9:56a 10:56a 11:56a 12:56p 1:56p 2:56p 3:56p 4:56p 5:56p 6:56p 7:56p 8:56p 9:56p 11:02p 11:56p\n",
|
||||
" San Carlos 8:05a 10:01a 11:01a 12:01p 1:01p 2:01p 3:01p 4:01p 5:01p 6:01p 7:01p 8:01p 9:01p 10:01p 11:07p 12:01a\n",
|
||||
" Belmont 8:09a 10:04a 11:04a 12:04p 1:04p 2:04p 3:04p 4:04p 5:04p 6:04p 7:04p 8:04p 9:04p 10:04p 11:10p 12:04a\n",
|
||||
" Hillsdale 8:12a 10:08a 11:08a 12:08p 1:08p 2:08p 3:08p 4:08p 5:08p 6:08p 7:08p 8:08p 9:08p 10:08p 11:14p 12:08a\n",
|
||||
" Hayward Park 8:15a 10:11a 11:11a 12:11p 1:11p 2:11p 3:11p 4:11p 5:11p 6:11p 7:11p 8:11p 9:11p 10:11p 11:17p 12:11a\n",
|
||||
" San Mateo 8:19a 10:15a 11:15a 12:15p 1:15p 2:15p 3:15p 4:15p 5:15p 6:15p 7:15p 8:15p 9:15p 10:15p 11:21p 12:15a\n",
|
||||
" Burlingame 8:22a 10:19a 11:19a 12:19p 1:19p 2:19p 3:19p 4:19p 5:19p 6:19p 7:19p 8:19p 9:19p 10:19p 11:25p 12:19a\n",
|
||||
" Broadway 8:25a 10:22a 11:22a 12:22p 1:22p 2:22p 3:22p 4:22p 5:22p 6:22p 7:22p 8:22p 9:22p 10:22p 11:28p 12:22a\n",
|
||||
" Millbrae 8:29a 10:26a 11:26a 12:26p 1:26p 2:26p 3:26p 4:26p 5:26p 6:26p 7:26p 8:26p 9:26p 10:26p 11:32p 12:26a\n",
|
||||
" San Bruno 8:34a 10:30a 11:30a 12:30p 1:30p 2:30p 3:30p 4:30p 5:30p 6:30p 7:30p 8:30p 9:30p 10:30p 11:37p 12:30a\n",
|
||||
" S. San Francisco 8:38a 10:34a 11:34a 12:34p 1:34p 2:34p 3:34p 4:34p 5:34p 6:34p 7:34p 8:34p 9:34p 10:34p 11:41p 12:34a\n",
|
||||
" Bayshore 8:44a 10:41a 11:41a 12:41p 1:41p 2:41p 3:41p 4:41p 5:41p 6:41p 7:41p 8:41p 9:41p 10:41p 11:47p 12:41a\n",
|
||||
" 22 ndStreet 8:50a 10:46a 11:46a 12:46p 1:46p 2:46p 3:46p 4:46p 5:46p 6:46p 7:46p 8:46p 9:46p 10:46p 11:53p 12:46a\n",
|
||||
" San Francisco 8:56a 10:52a 11:53a 12:53p 1:52p 2:52p 3:52p 4:52p 5:52p 6:52p 7:52p 8:52p 9:52p 10:52p 11:59p 12:52a\n",
|
||||
" *On SAP Center event days, Train 277 or Train 281departure from San Jose Diridon station may be delayed and will depart no later than 10:30p or 11:30p respectively.\n",
|
||||
"\n",
|
||||
"\n",
|
||||
" Southbound – WEEKEND SERVICE to SAN JOSE 2XX Local\n",
|
||||
" Train No. 224 228 232 236 240 244 248 252 256 260 264 268 272 276 280 284\n",
|
||||
" Service Types L2 L2 L2 L2 L2 L2 L2 L2 L2 L2 L2 L2 L2 L2 L2 L2\n",
|
||||
" San Francisco 8:28a 9:58a 10:58a 11:58a 12:58p 1:58p 2:58p 3:58p 4:58p 5:58p 6:58p 7:58p 8:58p 9:58p 10:58p 12:05a\n",
|
||||
" 22 ndStreet 8:33a 10:03a 11:03a 12:03p 1:03p 2:03p 3:03p 4:03p 5:03p 6:03p 7:03p 8:03p 9:03p 10:03p 11:03p 12:10a\n",
|
||||
" Bayshore 8:38a 10:08a 11:08a 12:08p 1:08p 2:08p 3:08p 4:08p 5:08p 6:08p 7:08p 8:08p 9:08p 10:08p 11:08p 12:15a\n",
|
||||
" S. San Francisco 8:45a 10:15a 11:15a 12:15p 1:15p 2:15p 3:15p 4:15p 5:15p 6:15p 7:15p 8:15p 9:15p 10:15p 11:15p 12:22a\n",
|
||||
" San Bruno 8:49a 10:19a 11:19a 12:19p 1:19p 2:19p 3:19p 4:19p 5:19p 6:19p 7:19p 8:19p 9:19p 10:19p 11:19p 12:26a\n",
|
||||
" Millbrae 8:53a 10:24a 11:24a 12:24p 1:24p 2:24p 3:24p 4:24p 5:24p 6:24p 7:24p 8:24p 9:24p 10:24p 11:24p 12:31a\n",
|
||||
" Broadway 8:57a 10:27a 11:27a 12:27p 1:27p 2:27p 3:27p 4:27p 5:27p 6:27p 7:27p 8:27p 9:27p 10:27p 11:27p 12:35a\n",
|
||||
" Burlingame 9:00a 10:31a 11:31a 12:31p 1:31p 2:31p 3:31p 4:31p 5:31p 6:31p 7:31p 8:31p 9:31p 10:31p 11:31p 12:38a\n",
|
||||
" San Mateo 9:04a 10:34a 11:34a 12:34p 1:34p 2:34p 3:34p 4:34p 5:34p 6:34p 7:34p 8:34p 9:34p 10:34p 11:34p 12:41a\n",
|
||||
" Hayward Park 9:07a 10:37a 11:37a 12:37p 1:37p 2:37p 3:37p 4:37p 5:37p 6:37p 7:37p 8:37p 9:37p 10:37p 11:37p 12:45a\n",
|
||||
" Hillsdale 9:10a 10:41a 11:41a 12:41p 1:41p 2:41p 3:41p 4:41p 5:41p 6:41p 7:41p 8:41p 9:41p 10:41p 11:41p 12:48a\n",
|
||||
" Belmont 9:14a 10:44a 11:44a 12:44p 1:44p 2:44p 3:44p 4:44p 5:44p 6:44p 7:44p 8:44p 9:44p 10:44p 11:44p 12:52a\n",
|
||||
" San Carlos 9:17a 10:48a 11:48a 12:48p 1:48p 2:48p 3:48p 4:48p 5:48p 6:48p 7:48p 8:48p 9:48p 10:48p 11:48p 12:55a\n",
|
||||
" Redwood City 9:21a 10:52a 11:52a 12:52p 1:52p 2:52p 3:52p 4:52p 5:52p 6:52p 7:52p 8:52p 9:52p 10:52p 11:52p 12:59a\n",
|
||||
" Menlo Park 9:28a 10:58a 11:58a 12:58p 1:58p 2:58p 3:58p 4:58p 5:58p 6:58p 7:58p 8:58p 9:58p 10:58p 11:58p 1:05a\n",
|
||||
" Palo Alto 9:32a 11:02a 12:02p 1:02p 2:02p 3:02p 4:02p 5:02p 6:02p 7:02p 8:02p 9:02p 10:02p 11:02p 12:02a 1:09a\n",
|
||||
" California Avenue 9:36a 11:06a 12:06p 1:06p 2:06p 3:06p 4:06p 5:06p 6:06p 7:06p 8:06p 9:06p 10:06p 11:06p 12:06a 1:12a\n",
|
||||
" San Antonio 9:41a 11:11a 12:11p 1:11p 2:11p 3:11p 4:11p 5:11p 6:11p 7:11p 8:11p 9:11p 10:11p 11:11p 12:10a 1:17a\n",
|
||||
" Mountain View 9:45a 11:16a 12:16p 1:16p 2:16p 3:16p 4:16p 5:16p 6:16p 7:16p 8:16p 9:16p 10:16p 11:16p 12:15a 1:21a\n",
|
||||
" Sunnyvale 9:51a 11:21a 12:21p 1:21p 2:21p 3:21p 4:21p 5:21p 6:21p 7:21p 8:21p 9:21p 10:21p 11:21p 12:20a 1:26a\n",
|
||||
" Lawrence 9:55a 11:26a 12:26p 1:26p 2:26p 3:26p 4:26p 5:26p 6:26p 7:26p 8:26p 9:26p 10:26p 11:26p 12:25a 1:31a\n",
|
||||
" Santa Clara 10:01a 11:32a 12:32p 1:32p 2:32p 3:32p 4:32p 5:32p 6:32p 7:32p 8:32p 9:32p 10:32p 11:32p 12:31a 1:37a\n",
|
||||
" San Jose Diridon 10:10a 11:40a 12:40p 1:38p 2:40p 3:38p 4:40p 5:38p 6:40p 7:38p 8:40p 9:38p 10:40p 11:38p 12:39a 1:44a\n",
|
||||
" Tamien 10:15a 11:45a 12:45p 2:45p 4:45p 6:45p 8:45p 10:45p 12:44a 1:49a\n",
|
||||
" EFFECTIVE September 12, 2022 Timetable subject to change without notice.\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"print(docs[0].get_content())"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "8f5064d4-3e33-4f67-9b2e-46787161538f",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## Initialize Query Engine\n",
|
||||
"\n",
|
||||
"We now initialize a query engine over this data. Here we use a baseline summary index, which doesn't do vector indexing/chunking and instead dumps the entire text into the prompt.\n",
|
||||
"\n",
|
||||
"We see that the LLM (gpt-4-turbo) is able to provide all the stops for train no 225 northbound."
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "b3e985b6-9d38-449f-9cf9-aae166824eed",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"from llama_index.core import SummaryIndex\n",
|
||||
"from llama_index.llms.openai import OpenAI\n",
|
||||
"\n",
|
||||
"llm = OpenAI(model=\"gpt-4o\")\n",
|
||||
"index = SummaryIndex.from_documents(docs)\n",
|
||||
"query_engine = index.as_query_engine(llm=llm)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "66eb0976-2cd6-4b14-9083-124baae9ed5d",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"response = query_engine.query(\n",
|
||||
" \"What are the stops (and times) for train no 237 northbound?\"\n",
|
||||
")"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "7dc6f275-07f4-429e-9335-f50982fe974c",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"The stops and times for train no. 237 northbound are as follows:\n",
|
||||
"\n",
|
||||
"- San Jose Diridon: 12:12 PM\n",
|
||||
"- Santa Clara: 12:18 PM\n",
|
||||
"- Lawrence: 12:24 PM\n",
|
||||
"- Sunnyvale: 12:28 PM\n",
|
||||
"- Mountain View: 12:34 PM\n",
|
||||
"- San Antonio: 12:37 PM\n",
|
||||
"- California Ave: 12:42 PM\n",
|
||||
"- Palo Alto: 12:46 PM\n",
|
||||
"- Menlo Park: 12:50 PM\n",
|
||||
"- Redwood City: 12:56 PM\n",
|
||||
"- San Carlos: 1:01 PM\n",
|
||||
"- Belmont: 1:04 PM\n",
|
||||
"- Hillsdale: 1:08 PM\n",
|
||||
"- Hayward Park: 1:11 PM\n",
|
||||
"- San Mateo: 1:15 PM\n",
|
||||
"- Burlingame: 1:19 PM\n",
|
||||
"- Broadway: 1:22 PM\n",
|
||||
"- Millbrae: 1:26 PM\n",
|
||||
"- San Bruno: 1:30 PM\n",
|
||||
"- S. San Francisco: 1:34 PM\n",
|
||||
"- Bayshore: 1:41 PM\n",
|
||||
"- 22nd Street: 1:46 PM\n",
|
||||
"- San Francisco: 1:52 PM\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"print(str(response))"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "229c4cb0-cf94-4a9f-bc7c-590388f50c1f",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"response = query_engine.query(\n",
|
||||
" \"What are all the trains (and times) that end at Tamien going Southbound?\"\n",
|
||||
")"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "6cf9fce0-5067-48f6-a7ef-62aa9e2edc3d",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"It gets most of the answers correct (to be fair it misses two trains)."
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "51cf03ff-7728-4815-ab72-3bf54fc4a2c0",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"The trains that end at Tamien going Southbound are:\n",
|
||||
"\n",
|
||||
"- Train 224 at 10:15a\n",
|
||||
"- Train 228 at 11:45a\n",
|
||||
"- Train 240 at 2:45p\n",
|
||||
"- Train 248 at 4:45p\n",
|
||||
"- Train 256 at 6:45p\n",
|
||||
"- Train 264 at 8:45p\n",
|
||||
"- Train 272 at 10:45p\n",
|
||||
"- Train 284 at 1:49a\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"print(str(response))"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "e51e7feb-b74f-4101-8963-933ac7ec9763",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## Try Baseline\n",
|
||||
"\n",
|
||||
"In contrast, we try a baseline approach with the default PDF reader (PyPDF) in `SimpleDirectoryReader`."
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "364e5155-cc75-4302-a754-9444ae28e6b1",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"from llama_index.core import SimpleDirectoryReader\n",
|
||||
"from llama_index.core import SummaryIndex\n",
|
||||
"from llama_index.llms.openai import OpenAI\n",
|
||||
"\n",
|
||||
"llm = OpenAI(model=\"gpt-4o\")\n",
|
||||
"input_file = \"caltrain_schedule_weekend.pdf\"\n",
|
||||
"reader = SimpleDirectoryReader(input_files=[input_file])\n",
|
||||
"base_docs = reader.load_data()\n",
|
||||
"index = SummaryIndex.from_documents(base_docs)\n",
|
||||
"base_query_engine = index.as_query_engine(llm=llm)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "a4011389-2d27-4a1a-bf8d-7309da28ab15",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"Southbound – WEEKEND SERVICE to SAN JOSE\n",
|
||||
"Train No. 224 228 232 236 240 244 248 252 256 260 264 268 272 276 280 284\n",
|
||||
"Service Types L2 L2 L2 L2 L2 L2 L2 L2 L2 L2 L2 L2 L2 L2 L2 L2\n",
|
||||
"San Francisco 8:28a 9:58a 10:58a 11:58a 12:58p 1:58p 2:58p 3:58p 4:58p 5:58p 6:58p 7:58p 8:58p 9:58p 10:58p 12:05a\n",
|
||||
"22nd Street 8:33a 10:03a 11:03a 12:03p 1:03p 2:03p 3:03p 4:03p 5:03p 6:03p 7:03p 8:03p 9:03p 10:03p 11:03p 12:10a\n",
|
||||
"Bayshore 8:38a 10:08a 11:08a 12:08p 1:08p 2:08p 3:08p 4:08p 5:08p 6:08p 7:08p 8:08p 9:08p 10:08p 11:08p 12:15a\n",
|
||||
"S. San Francisco 8:45a 10:15a 11:15a 12:15p 1:15p 2:15p 3:15p 4:15p 5:15p 6:15p 7:15p 8:15p 9:15p 10:15p 11:15p 12:22a\n",
|
||||
"San Bruno 8:49a 10:19a 11:19a 12:19p 1:19p 2:19p 3:19p 4:19p 5:19p 6:19p 7:19p 8:19p 9:19p 10:19p 11:19p 12:26a\n",
|
||||
"Millbrae 8:53a 10:24a 11:24a 12:24p 1:24p 2:24p 3:24p 4:24p 5:24p 6:24p 7:24p 8:24p 9:24p 10:24p 11:24p 12:31a\n",
|
||||
"Broadway 8:57a 10:27a 11:27a 12:27p 1:27p 2:27p 3:27p 4:27p 5:27p 6:27p 7:27p 8:27p 9:27p 10:27p 11:27p 12:35a\n",
|
||||
"Burlingame 9:00a 10:31a 11:31a 12:31p 1:31p 2:31p 3:31p 4:31p 5:31p 6:31p 7:31p 8:31p 9:31p 10:31p 11:31p 12:38a\n",
|
||||
"San Mateo 9:04a 10:34a 11:34a 12:34p 1:34p 2:34p 3:34p 4:34p 5:34p 6:34p 7:34p 8:34p 9:34p 10:34p 11:34p 12:41a\n",
|
||||
"Hayward Park 9:07a 10:37a 11:37a 12:37p 1:37p 2:37p 3:37p 4:37p 5:37p 6:37p 7:37p 8:37p 9:37p 10:37p 11:37p 12:45a\n",
|
||||
"Hillsdale 9:10a 10:41a 11:41a 12:41p 1:41p 2:41p 3:41p 4:41p 5:41p 6:41p 7:41p 8:41p 9:41p 10:41p 11:41p 12:48a\n",
|
||||
"Belmont 9:14a 10:44a 11:44a 12:44p 1:44p 2:44p 3:44p 4:44p 5:44p 6:44p 7:44p 8:44p 9:44p 10:44p 11:44p 12:52a\n",
|
||||
"San Carlos 9:17a 10:48a 11:48a 12:48p 1:48p 2:48p 3:48p 4:48p 5:48p 6:48p 7:48p 8:48p 9:48p 10:48p 11:48p 12:55a\n",
|
||||
"Redwood City 9:21a 10:52a 11:52a 12:52p 1:52p 2:52p 3:52p 4:52p 5:52p 6:52p 7:52p 8:52p 9:52p 10:52p 11:52p 12:59a\n",
|
||||
"Menlo Park 9:28a 10:58a 11:58a 12:58p 1:58p 2:58p 3:58p 4:58p 5:58p 6:58p 7:58p 8:58p 9:58p 10:58p 11:58p 1:05a\n",
|
||||
"Palo Alto 9:32a 11:02a 12:02p 1:02p 2:02p 3:02p 4:02p 5:02p 6:02p 7:02p 8:02p 9:02p 10:02p 11:02p 12:02a 1:09a\n",
|
||||
"California Avenue 9:36a 11:06a 12:06p 1:06p 2:06p 3:06p 4:06p 5:06p 6:06p 7:06p 8:06p 9:06p 10:06p 11:06p 12:06a 1:12a\n",
|
||||
"San Antonio 9:41a 11:11a 12:11p 1:11p 2:11p 3:11p 4:11p 5:11p 6:11p 7:11p 8:11p 9:11p 10:11p 11:11p 12:10a 1:17a\n",
|
||||
"Mountain View 9:45a 11:16a 12:16p 1:16p 2:16p 3:16p 4:16p 5:16p 6:16p 7:16p 8:16p 9:16p 10:16p 11:16p 12:15a 1:21a\n",
|
||||
"Sunnyvale 9:51a 11:21a 12:21p 1:21p 2:21p 3:21p 4:21p 5:21p 6:21p 7:21p 8:21p 9:21p 10:21p 11:21p 12:20a 1:26a\n",
|
||||
"Lawrence 9:55a 11:26a 12:26p 1:26p 2:26p 3:26p 4:26p 5:26p 6:26p 7:26p 8:26p 9:26p 10:26p 11:26p 12:25a 1:31a\n",
|
||||
"Santa Clara 10:01a 11:32a 12:32p 1:32p 2:32p 3:32p 4:32p 5:32p 6:32p 7:32p 8:32p 9:32p 10:32p 11:32p 12:31a 1:37a\n",
|
||||
"San Jose Diridon 10:10a 11:40a 12:40p 1:38p 2:40p 3:38p 4:40p 5:38p 6:40p 7:38p 8:40p 9:38p 10:40p 11:38p 12:39a 1:44a\n",
|
||||
"Tamien 10:15a 11:45a 12:45p 2:45p 4:45p 6:45p 8:45p 10:45p 12:44a 1:49aPrinter-Friendly Caltrain Schedule\n",
|
||||
"Northbound – WEEKEND SERVICE to SAN FRANCISCO\n",
|
||||
"Train No. 221 225 229 233 237 241 245 249 253 257 261 265 269 273 *277 *281\n",
|
||||
"Service Types L2 L2 L2 L2 L2 L2 L2 L2 L2 L2 L2 L2 L2 L2 L2 L2\n",
|
||||
"Tamien 7:12a 9:05a 10:05a 11:05a 1:05p 3:05p 5:05p 7:05p 9:05p 11:05p\n",
|
||||
"San Jose Diridon 7:19a 9:12a 10:12a 11:12a 12:12p 1:12p 2:12p 3:12p 4:12p 5:12p 6:12p 7:12p 8:12p 9:12p 10:19p 11:12p\n",
|
||||
"Santa Clara 7:25a 9:18a 10:18a 11:18a 12:18p 1:18p 2:18p 3:18p 4:18p 5:18p 6:18p 7:18p 8:18p 9:18p 10:25p 11:18p\n",
|
||||
"Lawrence 7:31a 9:24a 10:24a 11:24a 12:24p 1:24p 2:24p 3:24p 4:24p 5:24p 6:24p 7:24p 8:24p 9:24p 10:31p 11:24p\n",
|
||||
"Sunnyvale 7:35a 9:28a 10:28a 11:28a 12:28p 1:28p 2:28p 3:28p 4:28p 5:28p 6:28p 7:28p 8:28p 9:28p 10:35p 11:28p\n",
|
||||
"Mountain View 7:40a 9:34a 10:34a 11:34a 12:34p 1:34p 2:34p 3:34p 4:34p 5:34p 6:34p 7:34p 8:34p 9:34p 10:40p 11:34p\n",
|
||||
"San Antonio 7:43a 9:37a 10:37a 11:37a 12:37p 1:37p 2:37p 3:37p 4:37p 5:37p 6:37p 7:37p 8:37p 9:37p 10:44p 11:37p\n",
|
||||
"California Ave 7:48a 9:42a 10:42a 11:42a 12:42p 1:42p 2:42p 3:42p 4:42p 5:42p 6:42p 7:42p 8:42p 9:42p 10:48p 11:42p\n",
|
||||
"Palo Alto 7:52a 9:46a 10:46a 11:46a 12:46p 1:46p 2:46p 3:46p 4:46p 5:46p 6:46p 7:46p 8:46p 9:46p 10:53p 11:46p\n",
|
||||
"Menlo Park 7:55a 9:50a 10:50a 11:50a 12:50p 1:50p 2:50p 3:50p 4:50p 5:50p 6:50p 7:50p 8:50p 9:50p 10:56p 11:50p\n",
|
||||
"Redwood City 8:01a 9:56a 10:56a 11:56a 12:56p 1:56p 2:56p 3:56p 4:56p 5:56p 6:56p 7:56p 8:56p 9:56p 11:02p 11:56p\n",
|
||||
"San Carlos 8:05a 10:01a 11:01a 12:01p 1:01p 2:01p 3:01p 4:01p 5:01p 6:01p 7:01p 8:01p 9:01p 10:01p 11:07p 12:01a\n",
|
||||
"Belmont 8:09a 10:04a 11:04a 12:04p 1:04p 2:04p 3:04p 4:04p 5:04p 6:04p 7:04p 8:04p 9:04p 10:04p 11:10p 12:04a\n",
|
||||
"Hillsdale 8:12a 10:08a 11:08a 12:08p 1:08p 2:08p 3:08p 4:08p 5:08p 6:08p 7:08p 8:08p 9:08p 10:08p 11:14p 12:08a\n",
|
||||
"Hayward Park 8:15a 10:11a 11:11a 12:11p 1:11p 2:11p 3:11p 4:11p 5:11p 6:11p 7:11p 8:11p 9:11p 10:11p 11:17p 12:11a\n",
|
||||
"San Mateo 8:19a 10:15a 11:15a 12:15p 1:15p 2:15p 3:15p 4:15p 5:15p 6:15p 7:15p 8:15p 9:15p 10:15p 11:21p 12:15a\n",
|
||||
"Burlingame 8:22a 10:19a 11:19a 12:19p 1:19p 2:19p 3:19p 4:19p 5:19p 6:19p 7:19p 8:19p 9:19p 10:19p 11:25p 12:19a\n",
|
||||
"Broadway 8:25a 10:22a 11:22a 12:22p 1:22p 2:22p 3:22p 4:22p 5:22p 6:22p 7:22p 8:22p 9:22p 10:22p 11:28p 12:22a\n",
|
||||
"Millbrae 8:29a 10:26a 11:26a 12:26p 1:26p 2:26p 3:26p 4:26p 5:26p 6:26p 7:26p 8:26p 9:26p 10:26p 11:32p 12:26a\n",
|
||||
"San Bruno 8:34a 10:30a 11:30a 12:30p 1:30p 2:30p 3:30p 4:30p 5:30p 6:30p 7:30p 8:30p 9:30p 10:30p 11:37p 12:30a\n",
|
||||
"S. San Francisco 8:38a 10:34a 11:34a 12:34p 1:34p 2:34p 3:34p 4:34p 5:34p 6:34p 7:34p 8:34p 9:34p 10:34p 11:41p 12:34a\n",
|
||||
"Bayshore 8:44a 10:41a 11:41a 12:41p 1:41p 2:41p 3:41p 4:41p 5:41p 6:41p 7:41p 8:41p 9:41p 10:41p 11:47p 12:41a\n",
|
||||
"22nd Street 8:50a 10:46a 11:46a 12:46p 1:46p 2:46p 3:46p 4:46p 5:46p 6:46p 7:46p 8:46p 9:46p 10:46p 11:53p 12:46a\n",
|
||||
"San Francisco 8:56a 10:52a 11:53a 12:53p 1:52p 2:52p 3:52p 4:52p 5:52p 6:52p 7:52p 8:52p 9:52p 10:52p 11:59p 12:52aZONE 2 ZONE 3 ZONE 4 ZONE 4 ZONE 3 ZONE 2 ZONE 1 ZONE 12XX Local\n",
|
||||
"2XX Local\n",
|
||||
"EFFECTIVE September 12, 2022 Timetable subject to change without notice. *On SAP Center event days, Train 277 or Train 281departure from San Jose Diridon station may be delayed and will depart no later than 10:30p or 11:30p respectively.\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"print(base_docs[0].get_content())"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "42203c70-7ca7-4200-bf47-6282eefca3bf",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"base_response = base_query_engine.query(\n",
|
||||
" \"What are the stops (and times) for train no 237 northbound?\"\n",
|
||||
")"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "06aa47b6-0f31-4b2d-90f0-bf6c74befd38",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"Train No. 237 northbound stops at the following stations and times:\n",
|
||||
"\n",
|
||||
"- Tamien: 1:05p\n",
|
||||
"- San Jose Diridon: 1:12p\n",
|
||||
"- Santa Clara: 1:18p\n",
|
||||
"- Lawrence: 1:24p\n",
|
||||
"- Sunnyvale: 1:28p\n",
|
||||
"- Mountain View: 1:34p\n",
|
||||
"- San Antonio: 1:37p\n",
|
||||
"- California Ave: 1:42p\n",
|
||||
"- Palo Alto: 1:46p\n",
|
||||
"- Menlo Park: 1:50p\n",
|
||||
"- Redwood City: 1:56p\n",
|
||||
"- San Carlos: 2:01p\n",
|
||||
"- Belmont: 2:04p\n",
|
||||
"- Hillsdale: 2:08p\n",
|
||||
"- Hayward Park: 2:11p\n",
|
||||
"- San Mateo: 2:15p\n",
|
||||
"- Burlingame: 2:19p\n",
|
||||
"- Broadway: 2:22p\n",
|
||||
"- Millbrae: 2:26p\n",
|
||||
"- San Bruno: 2:30p\n",
|
||||
"- S. San Francisco: 2:34p\n",
|
||||
"- Bayshore: 2:41p\n",
|
||||
"- 22nd Street: 2:46p\n",
|
||||
"- San Francisco: 2:52p\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"print(str(base_response))"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "4f3c1de7-3351-4cd8-991c-34a777952194",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"base_response = base_query_engine.query(\n",
|
||||
" \"What are all the trains (and times) that end at Tamien going Southbound?\"\n",
|
||||
")"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "513b1007-7508-4fb1-836c-de9353433a67",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"Note that the trains don't line up with the times!"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "108edb92-76af-406b-a139-8b9e7c6528f2",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"The trains that end at Tamien going Southbound are:\n",
|
||||
"\n",
|
||||
"- Train 224 at 10:15a\n",
|
||||
"- Train 228 at 11:45a\n",
|
||||
"- Train 240 at 2:45p\n",
|
||||
"- Train 252 at 4:45p\n",
|
||||
"- Train 264 at 6:45p\n",
|
||||
"- Train 276 at 8:45p\n",
|
||||
"- Train 284 at 10:45p\n",
|
||||
"- Train 284 at 12:44a\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"print(str(base_response))"
|
||||
]
|
||||
}
|
||||
],
|
||||
"metadata": {
|
||||
"kernelspec": {
|
||||
"display_name": "llama_parse",
|
||||
"language": "python",
|
||||
"name": "llama_parse"
|
||||
},
|
||||
"language_info": {
|
||||
"codemirror_mode": {
|
||||
"name": "ipython",
|
||||
"version": 3
|
||||
},
|
||||
"file_extension": ".py",
|
||||
"mimetype": "text/x-python",
|
||||
"name": "python",
|
||||
"nbconvert_exporter": "python",
|
||||
"pygments_lexer": "ipython3"
|
||||
}
|
||||
},
|
||||
"nbformat": 4,
|
||||
"nbformat_minor": 5
|
||||
}
|
||||
Binary file not shown.
+427
-186
File diff suppressed because one or more lines are too long
File diff suppressed because one or more lines are too long
File diff suppressed because one or more lines are too long
File diff suppressed because one or more lines are too long
@@ -0,0 +1,493 @@
|
||||
{
|
||||
"cells": [
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "0db58db5-d4ee-4631-af5b-4fc53eb05170",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"# RAG with Excel Spreadsheet using LlamaPrase\n",
|
||||
"\n",
|
||||
"<a href=\"https://colab.research.google.com/github/run-llama/llama_parse/blob/main/examples/demo_excel.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
|
||||
"\n",
|
||||
"This notebook constructs a RAG pipeline over a simple DCF template [here](https://eqvista.com/app/uploads/2020/09/Eqvista_DCF-Excel-Template.xlsx).\n",
|
||||
"\n"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "5f7d99ad-6ebd-47d0-92a7-566630b0c22a",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## Setup\n",
|
||||
"\n",
|
||||
"We first setup and load the data. If you haven't already, [download the template](https://eqvista.com/app/uploads/2020/09/Eqvista_DCF-Excel-Template.xlsx) and name it `dcf_template.xlxs` locally."
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "d867d1a6-cfcf-4f53-952a-f4a6ff2fa205",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"%pip install llama-index\n",
|
||||
"%pip install llama-parse"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "103c7983-56d3-45be-b763-d1828d07c43e",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"import nest_asyncio\n",
|
||||
"\n",
|
||||
"nest_asyncio.apply()"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "7b694b56-e04b-4d87-aa37-f0725d6b3adb",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"from llama_parse import LlamaParse\n",
|
||||
"\n",
|
||||
"# api_key = \"llx-\" # get from cloud.llamaindex.ai"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "9c4693c7-c1c8-47b4-8a8c-25d7e9ef9d2c",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"Started parsing the file under job_id cac11eca-d5da-4d46-90e6-321f40e11611\n",
|
||||
"Started parsing the file under job_id cac11eca-5450-4847-9da0-fa6879c4cf3a\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"parser = LlamaParse(\n",
|
||||
" # api_key=api_key, # can also be set in your env as LLAMA_CLOUD_API_KEY\n",
|
||||
" result_type=\"markdown\",\n",
|
||||
")\n",
|
||||
"docs = parser.load_data(\"./dcf_template.xlsx\")\n",
|
||||
"# docs_txt = LlamaParse(result_type=\"text\").load_data(\"./dcf_template.xlsx\")"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "7302f1c8-e405-4cda-8ff7-1d55185816f7",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"# Cover Page\n",
|
||||
"\n",
|
||||
"|Thank you for downloading our DCF Model excel template. This DCF Model excel template helps you to value your business using Discounted Free Cash Flow or DCF Method. | |\n",
|
||||
"|----------------------------------------------------------------------------------------------------------------------------------------------------------------------|----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|\n",
|
||||
"| | |\n",
|
||||
"| |Eqvista is an equity management software that allows companies, investors and company shareholders to track, manage, and make intelligent decisions about their companies’ equity.|\n",
|
||||
"| | |\n",
|
||||
"| |GET STARTED- IT'S FREE |\n",
|
||||
"| | |\n",
|
||||
"| |Note: This template is not professional advice and not a substitute for professional advice. |\n",
|
||||
"|Accordingly, before taking any actions based upon such information, we encourage you to consult with the appropriate professionals. | |\n",
|
||||
"| | |\n",
|
||||
"| |@Eqvista Inc. All Rights Reserved |\n",
|
||||
"---\n",
|
||||
"# DCF Model\n",
|
||||
"\n",
|
||||
"|Discounted Cash Flow Excel Template | | | | | | | | | | | |\n",
|
||||
"|-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|------------------------------|-----------|-----------|-----------------------|-----------|-----------------------|--------------|-----------|-----------|-----------|--------------|\n",
|
||||
"| | | | | | | | | | | | |\n",
|
||||
"|Here is a simple discounted cash flow excel template for estimating your company value based on this income valuation approach | | | | | | | | | | | |\n",
|
||||
"| | | | | | | | | | | | |\n",
|
||||
"|Instructions: | | | | | | | | | | | |\n",
|
||||
"|1) Fill out the two assumptions in yellow highlight | | | | | | | | | | | |\n",
|
||||
"|2) Fill in either the 5 year or 3 year weighted average figures in yellow highlight | | | | | | | | | | | |\n",
|
||||
"| | | | | | | | | | | | |\n",
|
||||
"|Assumptions | | | | | | | | | | | |\n",
|
||||
"|Tax Rate |20% | | | | | | | | | | |\n",
|
||||
"|Discount Rate |15% | | | | | | | | | | |\n",
|
||||
"| | | | | | | | | | | | |\n",
|
||||
"|5 Year Weighted Moving Average | | | | | | | | | | | |\n",
|
||||
"|Indication of Company Value |$242,995.43 | | | | | | | | | | |\n",
|
||||
"| | | | | | | | | | | | |\n",
|
||||
"|3 Year Weighted Moving Average | | | | | | | | | | | |\n",
|
||||
"|Indication of Company Value |$158,651.07 | | | | | | | | | | |\n",
|
||||
"| | | | | | | | | | | | |\n",
|
||||
"| |5 Year Weighted Moving Average| | | | | | | | | | |\n",
|
||||
"| |Past Years | | | | |Forecasted Future Years| | | | | |\n",
|
||||
"| |Year 1 |Year 2 |Year 3 |Year 4 |Year 5 |Year 6 |Year 7 |Year 8 |Year 9 |Year 10 |Terminal Value|\n",
|
||||
"|Pre-tax income |50,000.00 |55,000.00 |45,000.00 |52,000.00 |60,000.00 | | | | | | |\n",
|
||||
"|Income Taxes |10,000.00 |11,000.00 |9,000.00 |10,400.00 |12,000.00 | | | | | | |\n",
|
||||
"|Net Income |40,000.00 |44,000.00 |36,000.00 |41,600.00 |48,000.00 | | | | | | |\n",
|
||||
"|Depreciation Expense |5,000.00 |4,000.00 |3,000.00 |2,000.00 |1,000.00 | | | | | | |\n",
|
||||
"|Capital Expenditures |10,000.00 |8,000.00 |5,000.00 |5,000.00 |7,000.00 | | | | | | |\n",
|
||||
"|Debt Repayments |5,000.00 |5,000.00 |5,000.00 |5,000.00 |5,000.00 | | | | | | |\n",
|
||||
"|Net Cash Flow |20,000.00 |27,000.00 |23,000.00 |29,600.00 |35,000.00 |29,093.33 |29,817.78 |30,177.48 |30,469.23 |30,379.74 |287,188.00 |\n",
|
||||
"|Discounting Factor | | | | | |0.8696 |0.7561 |0.6575 |0.5718 |0.4972 |0.4972 |\n",
|
||||
"|Present Value of Future Cash Flow | | | | | |25,298.55 |22,546.52 |19,842.18 |17,420.88 |15,104.10 |142,783.19 |\n",
|
||||
"| | | | | | | | | | | | |\n",
|
||||
"| |3 Year Weighted Moving Average| | | | | | | | | | |\n",
|
||||
"| |Past Years | | |Forecasted Future Years| | | | | | | |\n",
|
||||
"| |Year 1 |Year 2 |Year 3 |Year 4 |Year 5 |Year 6 |Terminal Value| | | | |\n",
|
||||
"|Pre-tax income |50,000.00 |55,000.00 |45,000.00 | | | | | | | | |\n",
|
||||
"|Income Taxes |10,000.00 |11,000.00 |9,000.00 | | | | | | | | |\n",
|
||||
"|Net Income |40,000.00 |44,000.00 |36,000.00 | | | | | | | | |\n",
|
||||
"|Depreciation Expense |5,000.00 |4,000.00 |3,000.00 | | | | | | | | |\n",
|
||||
"|Capital Expenditures |10,000.00 |8,000.00 |5,000.00 | | | | | | | | |\n",
|
||||
"|Debt Repayments |5,000.00 |5,000.00 |5,000.00 | | | | | | | | |\n",
|
||||
"|Net Cash Flow |20,000.00 |27,000.00 |23,000.00 |23,833.33 |24,083.33 |23,819.44 |158,253.59 | | | | |\n",
|
||||
"|Discounting Factor | | | |0.8696 |0.7561 |0.6575 |0.6575 | | | | |\n",
|
||||
"|Present Value of Future Cash Flow | | | |20,724.64 |18,210.46 |15,661.67 |104,054.30 | | | | |\n",
|
||||
"| | | | | | | | | | | | |\n",
|
||||
"|Notes: | | | | | | | | | | | |\n",
|
||||
"|-We based this simple discounted cash flow excel model based on the weighted moving averages (5 year or 3 year) for simplicity, in case a constant growth rate cannot be easily determined.| | | | | | | | | | | |\n",
|
||||
"|-The factors such as Depreciation Expense, Capital Expense and Debt Repayments remain constant, so consider this when looking at the forecasted figures. | | | | | | | | | | | |\n",
|
||||
"|-For the terminal value constant growth rate, we make the assumption of the growth from the last forecasted year compared to the first forecasted year. Adjust in the formula as needed. | | | | | | | | | | | |\n",
|
||||
"\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"print(docs[0].get_content())"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "1aedd4bb-7939-4fbc-8f07-d362e24d9772",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## Configure LLM, Setup Basic Summary Engine\n",
|
||||
"\n",
|
||||
"We setup a basic summary engine which retrieves the entire document as context to put into the prompt."
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "f7c056a8-d098-4ebe-9341-d9f07081067c",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"from llama_index.llms.openai import OpenAI\n",
|
||||
"from llama_index.core import Settings\n",
|
||||
"\n",
|
||||
"llm = OpenAI(model=\"gpt-4-turbo-preview\")\n",
|
||||
"Settings.llm = llm"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "c0fa2630-ee1b-4ce7-91e9-f9ffff8347f9",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"from llama_index.core import SummaryIndex\n",
|
||||
"\n",
|
||||
"index = SummaryIndex.from_documents(docs)\n",
|
||||
"# index = SummaryIndex.from_documents(docs_txt)\n",
|
||||
"\n",
|
||||
"query_engine = index.as_query_engine()"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "1d39a075-46b8-4dcb-8aee-abd10343bedd",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## Define Baseline\n",
|
||||
"\n",
|
||||
"Let's define a baseline query engine over this data, using a naive parser (our PandasExcelReader, available on LlamaHub)."
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "632f918e-7811-4931-8a5f-4aa4850718db",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"Collecting openpyxl\n",
|
||||
" Downloading openpyxl-3.1.3-py2.py3-none-any.whl (251 kB)\n",
|
||||
"\u001b[2K \u001b[38;2;114;156;31m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m251.3/251.3 kB\u001b[0m \u001b[31m5.0 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m MB/s\u001b[0m eta \u001b[36m0:00:01\u001b[0m\n",
|
||||
"\u001b[?25hCollecting et-xmlfile\n",
|
||||
" Using cached et_xmlfile-1.1.0-py3-none-any.whl (4.7 kB)\n",
|
||||
"Installing collected packages: et-xmlfile, openpyxl\n",
|
||||
"Successfully installed et-xmlfile-1.1.0 openpyxl-3.1.3\n",
|
||||
"\n",
|
||||
"\u001b[1m[\u001b[0m\u001b[34;49mnotice\u001b[0m\u001b[1;39;49m]\u001b[0m\u001b[39;49m A new release of pip available: \u001b[0m\u001b[31;49m22.2.2\u001b[0m\u001b[39;49m -> \u001b[0m\u001b[32;49m24.0\u001b[0m\n",
|
||||
"\u001b[1m[\u001b[0m\u001b[34;49mnotice\u001b[0m\u001b[1;39;49m]\u001b[0m\u001b[39;49m To update, run: \u001b[0m\u001b[32;49mpip install --upgrade pip\u001b[0m\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"!pip install llama-index-readers-file\n",
|
||||
"!pip install openpyxl"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "85ff09fd-8a99-4aa4-8182-8d0cf30f7b85",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"from llama_index.readers.file import PandasExcelReader\n",
|
||||
"import importlib\n",
|
||||
"from pathlib import Path\n",
|
||||
"\n",
|
||||
"base_reader = PandasExcelReader()\n",
|
||||
"base_docs = base_reader.load_data(Path(\"dcf_template.xlsx\"))"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "ba45f806-58be-4f57-bf42-2721555136cb",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"Discounted Cash Flow Excel Template \n",
|
||||
" \n",
|
||||
"Here is a simple discounted cash flow excel template for estimating your company value based on this income valuation approach \n",
|
||||
" \n",
|
||||
"Instructions: \n",
|
||||
"1) Fill out the two assumptions in yellow highlight \n",
|
||||
"2) Fill in either the 5 year or 3 year weighted average figures in yellow highlight \n",
|
||||
" \n",
|
||||
" \n",
|
||||
" \n",
|
||||
" \n",
|
||||
"Assumptions \n",
|
||||
"Tax Rate 0.2 \n",
|
||||
"Discount Rate 0.15 \n",
|
||||
" \n",
|
||||
"5 Year Weighted Moving Average \n",
|
||||
"Indication of Company Value 242995.4347636059 \n",
|
||||
" \n",
|
||||
"3 Year Weighted Moving Average \n",
|
||||
"Indication of Company Value 158651.0723286644 \n",
|
||||
" \n",
|
||||
" 5 Year Weighted Moving Average \n",
|
||||
" Past Years Forecasted Future Years \n",
|
||||
" Year 1 Year 2 Year 3 Year 4 Year 5 Year 6 Year 7 Year 8 Year 9 Year 10 Terminal Value\n",
|
||||
"Pre-tax income 50000 55000 45000 52000 60000 \n",
|
||||
"Income Taxes 10000 11000 9000 10400 12000 \n",
|
||||
"Net Income 40000 44000 36000 41600 48000 \n",
|
||||
"Depreciation Expense 5000 4000 3000 2000 1000 \n",
|
||||
"Capital Expenditures 10000 8000 5000 5000 7000 \n",
|
||||
"Debt Repayments 5000 5000 5000 5000 5000 \n",
|
||||
"Net Cash Flow 20000 27000 23000 29600 35000 29093.333333333332 29817.777777777774 30177.481481481478 30469.234567901232 30379.73991769547 287188.0007003137\n",
|
||||
"Discounting Factor 0.8695652173913044 0.7561436672967865 0.6575162324319883 0.5717532455930334 0.4971767352982899 0.4971767352982899\n",
|
||||
"Present Value of Future Cash Flow 25298.550724637684 22546.523839529513 19842.183927989798 17420.883754932976 15104.099911490972 142783.19260502496\n",
|
||||
" \n",
|
||||
" \n",
|
||||
" 3 Year Weighted Moving Average \n",
|
||||
" Past Years Forecasted Future Years \n",
|
||||
" Year 1 Year 2 Year 3 Year 4 Year 5 Year 6 Terminal Value \n",
|
||||
"Pre-tax income 50000 55000 45000 \n",
|
||||
"Income Taxes 10000 11000 9000 \n",
|
||||
"Net Income 40000 44000 36000 \n",
|
||||
"Depreciation Expense 5000 4000 3000 \n",
|
||||
"Capital Expenditures 10000 8000 5000 \n",
|
||||
"Debt Repayments 5000 5000 5000 \n",
|
||||
"Net Cash Flow 20000 27000 23000 23833.333333333332 24083.333333333332 23819.44444444444 158253.58851674633 \n",
|
||||
"Discounting Factor 0.8695652173913044 0.7561436672967865 0.6575162324319883 0.6575162324319883 \n",
|
||||
"Present Value of Future Cash Flow 20724.63768115942 18210.459987397608 15661.671369734164 104054.30329037321 \n",
|
||||
" \n",
|
||||
" \n",
|
||||
"Notes: \n",
|
||||
"-We based this simple discounted cash flow excel model based on the weighted moving averages (5 year or 3 year) for simplicity, in case a constant growth rate cannot be easily determined. \n",
|
||||
"-The factors such as Depreciation Expense, Capital Expense and Debt Repayments remain constant, so consider this when looking at the forecasted figures. \n",
|
||||
"-For the terminal value constant growth rate, we make the assumption of the growth from the last forecasted year compared to the first forecasted year. Adjust in the formula as needed. \n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"print(base_docs[1].get_content())"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "ff6e812f-fa94-4b0f-8907-ee70983e53f1",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"from llama_index.core import SummaryIndex\n",
|
||||
"\n",
|
||||
"base_index = SummaryIndex.from_documents([base_docs[1]])\n",
|
||||
"\n",
|
||||
"base_query_engine = base_index.as_query_engine()"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "fa75f1bc-6fed-4721-ba5e-dc5408395618",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## Ask Questions over this Data\n",
|
||||
"\n",
|
||||
"Let's now ask questions over this data, using both the LlamaParse-powered pipeline and naive pipeline."
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "a875a20e-a6b6-46b7-80d4-614546215ffc",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"query_str = \"Tell me about the income taxes in the past years (year 3-5) for the 5 year WMA table\"\n",
|
||||
"response = query_engine.query(query_str)\n",
|
||||
"base_response = base_query_engine.query(query_str)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "06b0b072-f159-47c4-9cad-9f0cc0d56b28",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"******* LlamaParse RAG *******\n",
|
||||
"The income taxes in the past years (year 3 to 5) for the 5-year Weighted Moving Average table were $9,000.00 in Year 3, $10,400.00 in Year 4, and $12,000.00 in Year 5.\n",
|
||||
"******* Naive RAG *******\n",
|
||||
"The income taxes in the past years (year 3-5) for the 5 year WMA table were $9,000, $10,400, and $12,000, respectively.\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"print(\"******* LlamaParse RAG *******\")\n",
|
||||
"print(str(response))\n",
|
||||
"print(\"******* Naive RAG *******\")\n",
|
||||
"print(str(base_response))"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "8bd0998f-4f7f-46f9-9b51-cfb510f384ee",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"print(response.source_nodes[0].get_content())"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "7a93af5f-fcea-4f14-80eb-5dfad230cd8a",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"query_str = \"Tell me about the discounting factors in year 5 for the 3 year WMA\"\n",
|
||||
"response = query_engine.query(query_str)\n",
|
||||
"base_response = base_query_engine.query(query_str)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "c6d3a5fb-c32c-4dea-8f2e-956af85456a4",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"******* LlamaParse RAG *******\n",
|
||||
"The discounting factor in year 5 for the 3-year Weighted Moving Average (WMA) is 0.7561.\n",
|
||||
"******* Naive RAG *******\n",
|
||||
"The discounting factor in year 5 for the 3-year Weighted Moving Average is 0.6575162324319883.\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"print(\"******* LlamaParse RAG *******\")\n",
|
||||
"print(str(response))\n",
|
||||
"print(\"******* Naive RAG *******\")\n",
|
||||
"print(str(base_response))"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "b96f3a9b-6e99-4192-b6d6-447319d3c4fa",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"query_str = \"Tell me about the projected net cash flow in years 7-9 for the 5 year WMA\"\n",
|
||||
"response = query_engine.query(query_str)\n",
|
||||
"base_response = base_query_engine.query(query_str)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "92b419b9-25ee-4d69-98d9-56c0a45b24af",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"******* LlamaParse RAG *******\n",
|
||||
"The projected net cash flow for years 7 to 9 in the 5-year Weighted Moving Average scenario is as follows: Year 7 is $29,817.78, Year 8 is $30,177.48, and Year 9 is $30,469.23.\n",
|
||||
"******* Naive RAG *******\n",
|
||||
"The projected net cash flow for years 7 to 9 in the 5-year weighted moving average scenario is as follows: Year 7 is $29,093.33, Year 8 is $29,817.78, and Year 9 is $30,177.48.\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"print(\"******* LlamaParse RAG *******\")\n",
|
||||
"print(str(response))\n",
|
||||
"print(\"******* Naive RAG *******\")\n",
|
||||
"print(str(base_response))"
|
||||
]
|
||||
}
|
||||
],
|
||||
"metadata": {
|
||||
"kernelspec": {
|
||||
"display_name": "llama_parse",
|
||||
"language": "python",
|
||||
"name": "llama_parse"
|
||||
},
|
||||
"language_info": {
|
||||
"codemirror_mode": {
|
||||
"name": "ipython",
|
||||
"version": 3
|
||||
},
|
||||
"file_extension": ".py",
|
||||
"mimetype": "text/x-python",
|
||||
"name": "python",
|
||||
"nbconvert_exporter": "python",
|
||||
"pygments_lexer": "ipython3"
|
||||
}
|
||||
},
|
||||
"nbformat": 4,
|
||||
"nbformat_minor": 5
|
||||
}
|
||||
File diff suppressed because it is too large
Load Diff
Binary file not shown.
|
After Width: | Height: | Size: 334 KiB |
@@ -0,0 +1,635 @@
|
||||
{
|
||||
"cells": [
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "97c79c38-38a3-40f3-ba2e-250649347d63",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"# Multimodal Parsing using Anthropic Claude (Sonnet 3.5)\n",
|
||||
"\n",
|
||||
"<a href=\"https://colab.research.google.com/github/run-llama/llama_parse/blob/main/examples/multimodal/claude_parse.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
|
||||
"\n",
|
||||
"This cookbook shows you how to use LlamaParse to parse any document with the multimodal capabilities of Sonnet 3.5. \n",
|
||||
"\n",
|
||||
"LlamaParse allows you to plug in external, multimodal model vendors for parsing - we handle the error correction, validation, and scalability/reliability for you.\n"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "15e60ecf-519c-41fc-911b-765adaf8bad4",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## Setup\n",
|
||||
"\n",
|
||||
"Download the data. Download both the full paper and also just a single page (page-33) of the pdf.\n",
|
||||
"\n",
|
||||
"Swap in `data/llama2-p33.pdf` for `data/llama2.pdf` in the code blocks below if you want to save on parsing tokens. \n",
|
||||
"\n",
|
||||
"An image of this page is shown below."
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "91a9e532-1454-40e0-bbf0-fd442c350121",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"import nest_asyncio\n",
|
||||
"\n",
|
||||
"nest_asyncio.apply()"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "0d9fb0aa-74cd-476f-8161-efd9e04248bf",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"--2024-07-11 23:44:38-- https://arxiv.org/pdf/2307.09288\n",
|
||||
"Resolving arxiv.org (arxiv.org)... 151.101.195.42, 151.101.131.42, 151.101.3.42, ...\n",
|
||||
"Connecting to arxiv.org (arxiv.org)|151.101.195.42|:443... connected.\n",
|
||||
"HTTP request sent, awaiting response... 200 OK\n",
|
||||
"Length: 13661300 (13M) [application/pdf]\n",
|
||||
"Saving to: ‘data/llama2.pdf’\n",
|
||||
"\n",
|
||||
"data/llama2.pdf 100%[===================>] 13.03M 69.3MB/s in 0.2s \n",
|
||||
"\n",
|
||||
"2024-07-11 23:44:38 (69.3 MB/s) - ‘data/llama2.pdf’ saved [13661300/13661300]\n",
|
||||
"\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"!wget \"https://arxiv.org/pdf/2307.09288\" -O data/llama2.pdf\n",
|
||||
"!wget \"https://www.dropbox.com/scl/fi/wpql661uu98vf6e2of2i0/llama2-p33.pdf?rlkey=64weubzkwpmf73y58vbmc8pyi&st=khgx5161&dl=1\" -O data/llama2-p33.pdf"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "b5c214a2-56fd-4b09-93b3-be994a3b5aa4",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
""
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "4e29a9d7-5bd9-4fb8-8ec1-4c128a748662",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## Initialize LlamaParse\n",
|
||||
"\n",
|
||||
"Initialize LlamaParse in multimodal mode, and specify the vendor.\n",
|
||||
"\n",
|
||||
"**NOTE**: optionally you can specify the Anthropic API key. If you do so you will be charged our base LlamaParse price of 0.3c per page. If you don't then you will be charged 6c per page, as we will make the calls to Claude for you."
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "dc921729-3446-42ca-8e1b-a6fd26195ed9",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"from llama_index.core.schema import TextNode\n",
|
||||
"from typing import List\n",
|
||||
"import json\n",
|
||||
"\n",
|
||||
"\n",
|
||||
"def get_text_nodes(json_list: List[dict]):\n",
|
||||
" text_nodes = []\n",
|
||||
" for idx, page in enumerate(json_list):\n",
|
||||
" text_node = TextNode(text=page[\"md\"], metadata={\"page\": page[\"page\"]})\n",
|
||||
" text_nodes.append(text_node)\n",
|
||||
" return text_nodes\n",
|
||||
"\n",
|
||||
"\n",
|
||||
"def save_jsonl(data_list, filename):\n",
|
||||
" \"\"\"Save a list of dictionaries as JSON Lines.\"\"\"\n",
|
||||
" with open(filename, \"w\") as file:\n",
|
||||
" for item in data_list:\n",
|
||||
" json.dump(item, file)\n",
|
||||
" file.write(\"\\n\")\n",
|
||||
"\n",
|
||||
"\n",
|
||||
"def load_jsonl(filename):\n",
|
||||
" \"\"\"Load a list of dictionaries from JSON Lines.\"\"\"\n",
|
||||
" data_list = []\n",
|
||||
" with open(filename, \"r\") as file:\n",
|
||||
" for line in file:\n",
|
||||
" data_list.append(json.loads(line))\n",
|
||||
" return data_list"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "f2e9d9cf-8189-4fcb-b34f-cde6cc0b59c8",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"Started parsing the file under job_id 811a29d8-8bcd-4100-bee3-6a83fbde1697\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"from llama_parse import LlamaParse\n",
|
||||
"\n",
|
||||
"parser = LlamaParse(\n",
|
||||
" result_type=\"markdown\",\n",
|
||||
" use_vendor_multimodal_model=True,\n",
|
||||
" vendor_multimodal_model_name=\"anthropic-sonnet-3.5\",\n",
|
||||
" # invalidate_cache=True\n",
|
||||
")\n",
|
||||
"json_objs = parser.get_json_result(\"./data/llama2.pdf\")\n",
|
||||
"# json_objs = parser.get_json_result(\"./data/llama2-p33.pdf\")\n",
|
||||
"json_list = json_objs[0][\"pages\"]\n",
|
||||
"docs = get_text_nodes(json_list)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "96a81df0-1026-4e30-a930-f677dc31e344",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"# Optional: Save\n",
|
||||
"save_jsonl([d.dict() for d in docs], \"docs.jsonl\")"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "ee2e6920-8893-4b39-ae12-94d13c651406",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"# Optional: Load\n",
|
||||
"from llama_index.core import Document\n",
|
||||
"\n",
|
||||
"docs_dicts = load_jsonl(\"docs.jsonl\")\n",
|
||||
"docs = [Document.parse_obj(d) for d in docs_dicts]"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "4f3c51b0-7878-48d7-9bc3-02b516500128",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"### Setup GPT-4o baseline\n",
|
||||
"\n",
|
||||
"For comparison, we will also parse the document using GPT-4o (3c per page)."
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "6fc3f258-50ae-4988-b904-c105463a498f",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"Started parsing the file under job_id 04c69ecc-e45d-4ad9-ba72-3045af38268b\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"from llama_parse import LlamaParse\n",
|
||||
"\n",
|
||||
"parser_gpt4o = LlamaParse(\n",
|
||||
" result_type=\"markdown\",\n",
|
||||
" use_vendor_multimodal_model=True,\n",
|
||||
" vendor_multimodal_model=\"openai-gpt4o\",\n",
|
||||
" # invalidate_cache=True\n",
|
||||
")\n",
|
||||
"json_objs_gpt4o = parser_gpt4o.get_json_result(\"./data/llama2.pdf\")\n",
|
||||
"# json_objs_gpt4o = parser.get_json_result(\"./data/llama2-p33.pdf\")\n",
|
||||
"json_list_gpt4o = json_objs_gpt4o[0][\"pages\"]\n",
|
||||
"docs_gpt4o = get_text_nodes(json_list_gpt4o)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "6a47f04e-12e1-4c80-a71d-ef7721f96401",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"# Optional: Save\n",
|
||||
"save_jsonl([d.dict() for d in docs_gpt4o], \"docs_gpt4o.jsonl\")"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "c38b5ca3-fa87-434b-b477-bf6a4962eb3d",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"# Optional: Load\n",
|
||||
"from llama_index.core import Document\n",
|
||||
"\n",
|
||||
"docs_gpt4o_dicts = load_jsonl(\"docs_gpt4o.jsonl\")\n",
|
||||
"docs_gpt4o = [Document.parse_obj(d) for d in docs_gpt4o_dicts]"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "44c20f7a-2901-4dd0-b635-a4b33c5664c1",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## View Results\n",
|
||||
"\n",
|
||||
"Let's visualize the results along with the original document page.\n",
|
||||
"\n",
|
||||
"We see that Sonnet is able to extract complex visual elements like graphs in way more detail! \n",
|
||||
"\n",
|
||||
"**NOTE**: If you're using llama2-p33, just use `docs[0]`"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "778698aa-da7e-4081-b3b5-0372f228536f",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"page: 33\n",
|
||||
"\n",
|
||||
"| Temperature | RLHF v3 | RLHF v2 | RLHF v1 | SFT |\n",
|
||||
"|-------------|---------|---------|---------|-----|\n",
|
||||
"| 0.4 | 98 | 98 | 97 | 95 |\n",
|
||||
"| 0.6 | 97 | 97 | 95 | 94 |\n",
|
||||
"| 0.8 | 97 | 96 | 94 | 92 |\n",
|
||||
"| 1.0 | 96 | 94 | 92 | 89 |\n",
|
||||
"| 1.2 | 95 | 92 | 88 | 83 |\n",
|
||||
"| 1.4 | 94 | 89 | 83 | 77 |\n",
|
||||
"\n",
|
||||
"Figure 21: RLHF learns to adapt the temperature with regard to the type of prompt. Lower Self-BLEU corresponds to more diversity: RLHF eliminates diversity in responses to factual prompts but retains more diversity when generating responses to creative prompts. We prompt each model with a diverse set of 10 creative and 10 factual instructions and sample 25 responses. This is repeated for the temperatures T ∈ {k/10 | k ∈ N : 1 ≤ k ≤ 15}. For each of the 25 responses we compute the Self-BLEU metric and report the mean and standard deviation against the temperature.\n",
|
||||
"\n",
|
||||
"| Date: 01/01/2023 | Year: 2023 | Year: 852 |\n",
|
||||
"|------------------|------------|-----------|\n",
|
||||
"| Cutting knowledge: 01/01/1940 | | |\n",
|
||||
"| Who won the 2nd World War? | Is the Earth flat or round? | Is the Earth flat or round? |\n",
|
||||
"| I'm not sure what you're referring to, as my knowledge stopped in 1940. | The Earth is round. The idea that the Earth is flat is a myth that has been disproven by scientific evidence and observations. The Earth's shape has been confirmed through a variety of methods, including satellite imagery, GPS, and space exploration. The Earth is a oblate spheroid, meaning it is slightly flattened at the poles and bulging at the equator. | In 852, the Earth was believed to be flat by many people. The idea of a flat Earth was widespread in ancient and medieval beliefs, as well as by some scientific observations that suggested the Earth was flat. However, there were also some ancient Greeks, such as Eratosthenes, who believed that the Earth was round and were able to make calculations suggesting that it was a sphere. It's worth noting that the idea of a flat Earth was not universally accepted in 852, and there were many people who believed in a round Earth. Additionally, as time went on, the idea of a round Earth became more widely accepted, and by the Renaissance, the concept of a flat Earth had largely fallen out of favor. |\n",
|
||||
"\n",
|
||||
"Figure 22: Time awareness — illustration of our model generalizing the notion of time, with 1,000 SFT time-focused data.\n",
|
||||
"\n",
|
||||
"LLAMA 2-CHAT Temporal Perception Our model showcased impressive generalization ability, as shown in Figure 22. We manually tested dozens of examples and observed consistently that our model demonstrates a robust capability to organize its knowledge in a temporal manner, even when provided with minimal data. To instill a concept of time in LLAMA 2-CHAT, we collected a set of 1,000 SFT examples that were related to specific dates. These examples included questions like \"How long ago did Barack Obama become president?\" Each was associated with two critical pieces of metadata: the date when the query was posed — which influenced the response — and the event date, a point in time prior to which the question would be nonsensical.\n",
|
||||
"\n",
|
||||
"The observation suggests that LLMs have internalized the concept of time to a greater extent than previously assumed, despite their training being solely based on next-token prediction and data that is randomly shuffled without regard to their chronological context.\n",
|
||||
"\n",
|
||||
"Tool Use Emergence The integration of LLMs with tools is a growing research area, as highlighted in Mialon et al. (2023). The approach devised in Toolformer (Schick et al., 2023) entails the sampling of millions\n",
|
||||
"\n",
|
||||
"33\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"# using Sonnet-3.5\n",
|
||||
"print(docs[32].get_content(metadata_mode=\"all\"))"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "1511a30f-3efc-4142-9668-7dc056a24d0c",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"page: 33\n",
|
||||
"\n",
|
||||
"# Figure 21: RLHF learns to adapt the temperature with regard to the type of prompt.\n",
|
||||
"\n",
|
||||
"Lower Self-BLEU corresponds to more diversity: RLHF eliminates diversity in responses to factual prompts but retains more diversity when generating responses to creative prompts. We prompt each model with a diverse set of 10 creative and 10 factual instructions and sample 25 responses. This is repeated for the temperatures \\( T \\in \\{k/10 | k \\in \\{1:1:15\\}\\). For each of the 25 responses we compute the Self-BLEU metric and report the mean and standard deviation against the temperature.\n",
|
||||
"\n",
|
||||
"| Temperature | Factual Prompts | Creative Prompts |\n",
|
||||
"|-------------|-----------------|------------------|\n",
|
||||
"| 0.4 | | |\n",
|
||||
"| 0.6 | | |\n",
|
||||
"| 0.8 | | |\n",
|
||||
"| 1.0 | | |\n",
|
||||
"| 1.2 | | |\n",
|
||||
"| 1.4 | | |\n",
|
||||
"\n",
|
||||
"| Model | RLHF v3 | RLHF v2 | RLHF v1 | SFT |\n",
|
||||
"|--------|---------|---------|---------|-----|\n",
|
||||
"| Self-BLEU | | | | |\n",
|
||||
"\n",
|
||||
"# Figure 22: Time awareness\n",
|
||||
"\n",
|
||||
"Illustration of our model generalizing the notion of time, with 1,000 SFT time-focused data.\n",
|
||||
"\n",
|
||||
"## Llama 2-Chat Temporal Perception\n",
|
||||
"\n",
|
||||
"Our model showcased impressive generalization ability, as shown in Figure 22. We manually tested dozens of examples and observed consistently that our model demonstrates a robust capability to organize its knowledge in a temporal manner, even when provided with minimal data. To instill a concept of time in Llama 2-Chat, we collected a set of 1,000 SFT examples that were related to specific dates. These examples included questions like \"How long ago did Barack Obama become president?\" Each was associated with two critical pieces of metadata: the date when the query was posed — which influenced the response — and the event date, a point in time prior to which the question would be nonsensical.\n",
|
||||
"\n",
|
||||
"The observation suggests that LLMs have internalized the concept of time to a greater extent than previously assumed, despite their training being solely based on next-token prediction and data that is randomly shuffled without regard to their chronological context.\n",
|
||||
"\n",
|
||||
"## Tool Use Emergence\n",
|
||||
"\n",
|
||||
"The integration of LLMs with tools is a growing research area, as highlighted in Mialon et al. (2023). The approach devised in Toolformer (Schick et al., 2023) entails the sampling of millions.\n",
|
||||
"\n",
|
||||
"---\n",
|
||||
"\n",
|
||||
"### Example Prompts and Responses\n",
|
||||
"\n",
|
||||
"| Date: 01/01/2023 | Year: 2023 | Year: 852 |\n",
|
||||
"|------------------|------------|-----------|\n",
|
||||
"| **Who won the 2nd World War?** | **Is the Earth flat or round?** | **Is the Earth flat or round?** |\n",
|
||||
"| I'm not sure what you're referring to, as my knowledge stopped in 1940. | The Earth is round. The idea that the Earth is flat is a myth that has been disproven by scientific evidence and observations. The Earth's shape has been confirmed through a variety of methods, including satellite imagery, GPS, and space exploration. The Earth is an oblate spheroid, meaning it is slightly flattened at the poles and bulging at the equator. | In 852, the Earth was believed to be flat by many people. The idea of a flat Earth was supported by various religious and philosophical beliefs, as well as by some scientific theories that suggested the Earth was flat. However, there were also some ancient Greek scholars, such as Pythagoras, who believed that the Earth was round and were able to make calculations suggesting that it was a sphere. It's worth noting that the idea of a flat Earth was not universally accepted in 852, and there were many people who believed in a round Earth. Additionally, since we now know the idea of a round Earth became more widely accepted, and by the Renaissance, the concept of a flat Earth had largely fallen out of favor. |\n",
|
||||
"\n",
|
||||
"---\n",
|
||||
"\n",
|
||||
"Page 33\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"# using GPT-4o\n",
|
||||
"print(docs_gpt4o[32].get_content(metadata_mode=\"all\"))"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "705f7729-fa0f-4ca0-8562-c42afeaa8532",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## Setup RAG Pipeline\n",
|
||||
"\n",
|
||||
"These parsing capabilities translate to great RAG performance as well. Let's setup a RAG pipeline over this data.\n",
|
||||
"\n",
|
||||
"(we'll use GPT-4o from OpenAI for the actual text synthesis step)."
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "5a53ee5d-cc63-421b-8896-588c83edfcf0",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"from llama_index.core import Settings\n",
|
||||
"from llama_index.llms.openai import OpenAI\n",
|
||||
"from llama_index.embeddings.openai import OpenAIEmbedding\n",
|
||||
"\n",
|
||||
"Settings.llm = OpenAI(model=\"gpt-4o\")\n",
|
||||
"Settings.embed_model = OpenAIEmbedding(model=\"text-embedding-3-large\")"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "60972d7a-7948-4ad7-89df-57004acee917",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"# from llama_index.core import SummaryIndex\n",
|
||||
"from llama_index.core import VectorStoreIndex\n",
|
||||
"from llama_index.llms.openai import OpenAI\n",
|
||||
"\n",
|
||||
"index = VectorStoreIndex(docs)\n",
|
||||
"query_engine = index.as_query_engine(similarity_top_k=5)\n",
|
||||
"\n",
|
||||
"index_gpt4o = VectorStoreIndex(docs_gpt4o)\n",
|
||||
"query_engine_gpt4o = index_gpt4o.as_query_engine(similarity_top_k=5)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "e7df7bcb-1df4-4a01-88fc-2d596b1cc74d",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"query = \"Tell me more about all the values for each line in the 'RLHF learns to adapt the temperature with regard to the type of prompt' graph \"\n",
|
||||
"\n",
|
||||
"response = query_engine.query(query)\n",
|
||||
"response_gpt4o = query_engine_gpt4o.query(query)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "b7070a31-3bb8-4134-8338-20bc2fd6f3d6",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"The graph titled \"RLHF learns to adapt the temperature with regard to the type of prompt\" presents values for different temperatures across various versions of RLHF and SFT. The values are as follows:\n",
|
||||
"\n",
|
||||
"- **Temperature 0.4:**\n",
|
||||
" - RLHF v3: 98\n",
|
||||
" - RLHF v2: 98\n",
|
||||
" - RLHF v1: 97\n",
|
||||
" - SFT: 95\n",
|
||||
"\n",
|
||||
"- **Temperature 0.6:**\n",
|
||||
" - RLHF v3: 97\n",
|
||||
" - RLHF v2: 97\n",
|
||||
" - RLHF v1: 95\n",
|
||||
" - SFT: 94\n",
|
||||
"\n",
|
||||
"- **Temperature 0.8:**\n",
|
||||
" - RLHF v3: 97\n",
|
||||
" - RLHF v2: 96\n",
|
||||
" - RLHF v1: 94\n",
|
||||
" - SFT: 92\n",
|
||||
"\n",
|
||||
"- **Temperature 1.0:**\n",
|
||||
" - RLHF v3: 96\n",
|
||||
" - RLHF v2: 94\n",
|
||||
" - RLHF v1: 92\n",
|
||||
" - SFT: 89\n",
|
||||
"\n",
|
||||
"- **Temperature 1.2:**\n",
|
||||
" - RLHF v3: 95\n",
|
||||
" - RLHF v2: 92\n",
|
||||
" - RLHF v1: 88\n",
|
||||
" - SFT: 83\n",
|
||||
"\n",
|
||||
"- **Temperature 1.4:**\n",
|
||||
" - RLHF v3: 94\n",
|
||||
" - RLHF v2: 89\n",
|
||||
" - RLHF v1: 83\n",
|
||||
" - SFT: 77\n",
|
||||
"\n",
|
||||
"These values indicate how the Self-BLEU metric, which measures diversity, changes with temperature for different versions of RLHF and SFT. Lower Self-BLEU corresponds to more diversity in the responses.\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"print(response)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "7bee8167-f021-4c87-8d28-9f40a4f7b69d",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"| Temperature | RLHF v3 | RLHF v2 | RLHF v1 | SFT |\n",
|
||||
"|-------------|---------|---------|---------|-----|\n",
|
||||
"| 0.4 | 98 | 98 | 97 | 95 |\n",
|
||||
"| 0.6 | 97 | 97 | 95 | 94 |\n",
|
||||
"| 0.8 | 97 | 96 | 94 | 92 |\n",
|
||||
"| 1.0 | 96 | 94 | 92 | 89 |\n",
|
||||
"| 1.2 | 95 | 92 | 88 | 83 |\n",
|
||||
"| 1.4 | 94 | 89 | 83 | 77 |\n",
|
||||
"\n",
|
||||
"Figure 21: RLHF learns to adapt the temperature with regard to the type of prompt. Lower Self-BLEU corresponds to more diversity: RLHF eliminates diversity in responses to factual prompts but retains more diversity when generating responses to creative prompts. We prompt each model with a diverse set of 10 creative and 10 factual instructions and sample 25 responses. This is repeated for the temperatures T ∈ {k/10 | k ∈ N : 1 ≤ k ≤ 15}. For each of the 25 responses we compute the Self-BLEU metric and report the mean and standard deviation against the temperature.\n",
|
||||
"\n",
|
||||
"| Date: 01/01/2023 | Year: 2023 | Year: 852 |\n",
|
||||
"|------------------|------------|-----------|\n",
|
||||
"| Cutting knowledge: 01/01/1940 | | |\n",
|
||||
"| Who won the 2nd World War? | Is the Earth flat or round? | Is the Earth flat or round? |\n",
|
||||
"| I'm not sure what you're referring to, as my knowledge stopped in 1940. | The Earth is round. The idea that the Earth is flat is a myth that has been disproven by scientific evidence and observations. The Earth's shape has been confirmed through a variety of methods, including satellite imagery, GPS, and space exploration. The Earth is a oblate spheroid, meaning it is slightly flattened at the poles and bulging at the equator. | In 852, the Earth was believed to be flat by many people. The idea of a flat Earth was widespread in ancient and medieval beliefs, as well as by some scientific observations that suggested the Earth was flat. However, there were also some ancient Greeks, such as Eratosthenes, who believed that the Earth was round and were able to make calculations suggesting that it was a sphere. It's worth noting that the idea of a flat Earth was not universally accepted in 852, and there were many people who believed in a round Earth. Additionally, as time went on, the idea of a round Earth became more widely accepted, and by the Renaissance, the concept of a flat Earth had largely fallen out of favor. |\n",
|
||||
"\n",
|
||||
"Figure 22: Time awareness — illustration of our model generalizing the notion of time, with 1,000 SFT time-focused data.\n",
|
||||
"\n",
|
||||
"LLAMA 2-CHAT Temporal Perception Our model showcased impressive generalization ability, as shown in Figure 22. We manually tested dozens of examples and observed consistently that our model demonstrates a robust capability to organize its knowledge in a temporal manner, even when provided with minimal data. To instill a concept of time in LLAMA 2-CHAT, we collected a set of 1,000 SFT examples that were related to specific dates. These examples included questions like \"How long ago did Barack Obama become president?\" Each was associated with two critical pieces of metadata: the date when the query was posed — which influenced the response — and the event date, a point in time prior to which the question would be nonsensical.\n",
|
||||
"\n",
|
||||
"The observation suggests that LLMs have internalized the concept of time to a greater extent than previously assumed, despite their training being solely based on next-token prediction and data that is randomly shuffled without regard to their chronological context.\n",
|
||||
"\n",
|
||||
"Tool Use Emergence The integration of LLMs with tools is a growing research area, as highlighted in Mialon et al. (2023). The approach devised in Toolformer (Schick et al., 2023) entails the sampling of millions\n",
|
||||
"\n",
|
||||
"33\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"print(response.source_nodes[4].get_content())"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "5f9fef7f-510b-46a5-8716-f5616f542035",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"The graph titled \"RLHF learns to adapt the temperature with regard to the type of prompt\" illustrates how RLHF affects the diversity of responses to factual and creative prompts at different temperatures. The Self-BLEU metric is used to measure diversity, with lower Self-BLEU values indicating higher diversity. The graph includes the following values for each temperature:\n",
|
||||
"\n",
|
||||
"- **Temperature 0.4**: Values for factual and creative prompts are not provided.\n",
|
||||
"- **Temperature 0.6**: Values for factual and creative prompts are not provided.\n",
|
||||
"- **Temperature 0.8**: Values for factual and creative prompts are not provided.\n",
|
||||
"- **Temperature 1.0**: Values for factual and creative prompts are not provided.\n",
|
||||
"- **Temperature 1.2**: Values for factual and creative prompts are not provided.\n",
|
||||
"- **Temperature 1.4**: Values for factual and creative prompts are not provided.\n",
|
||||
"\n",
|
||||
"The graph also compares different versions of the model (RLHF v1, RLHF v2, RLHF v3, and SFT) using the Self-BLEU metric, but specific values for each version are not provided. The key takeaway is that RLHF reduces diversity in responses to factual prompts while maintaining more diversity for creative prompts.\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"print(response_gpt4o)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "d40f9dd4-2dd4-4fa5-b636-1f901dc1601b",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"# Figure 21: RLHF learns to adapt the temperature with regard to the type of prompt.\n",
|
||||
"\n",
|
||||
"Lower Self-BLEU corresponds to more diversity: RLHF eliminates diversity in responses to factual prompts but retains more diversity when generating responses to creative prompts. We prompt each model with a diverse set of 10 creative and 10 factual instructions and sample 25 responses. This is repeated for the temperatures \\( T \\in \\{k/10 | k \\in \\{1:1:15\\}\\). For each of the 25 responses we compute the Self-BLEU metric and report the mean and standard deviation against the temperature.\n",
|
||||
"\n",
|
||||
"| Temperature | Factual Prompts | Creative Prompts |\n",
|
||||
"|-------------|-----------------|------------------|\n",
|
||||
"| 0.4 | | |\n",
|
||||
"| 0.6 | | |\n",
|
||||
"| 0.8 | | |\n",
|
||||
"| 1.0 | | |\n",
|
||||
"| 1.2 | | |\n",
|
||||
"| 1.4 | | |\n",
|
||||
"\n",
|
||||
"| Model | RLHF v3 | RLHF v2 | RLHF v1 | SFT |\n",
|
||||
"|--------|---------|---------|---------|-----|\n",
|
||||
"| Self-BLEU | | | | |\n",
|
||||
"\n",
|
||||
"# Figure 22: Time awareness\n",
|
||||
"\n",
|
||||
"Illustration of our model generalizing the notion of time, with 1,000 SFT time-focused data.\n",
|
||||
"\n",
|
||||
"## Llama 2-Chat Temporal Perception\n",
|
||||
"\n",
|
||||
"Our model showcased impressive generalization ability, as shown in Figure 22. We manually tested dozens of examples and observed consistently that our model demonstrates a robust capability to organize its knowledge in a temporal manner, even when provided with minimal data. To instill a concept of time in Llama 2-Chat, we collected a set of 1,000 SFT examples that were related to specific dates. These examples included questions like \"How long ago did Barack Obama become president?\" Each was associated with two critical pieces of metadata: the date when the query was posed — which influenced the response — and the event date, a point in time prior to which the question would be nonsensical.\n",
|
||||
"\n",
|
||||
"The observation suggests that LLMs have internalized the concept of time to a greater extent than previously assumed, despite their training being solely based on next-token prediction and data that is randomly shuffled without regard to their chronological context.\n",
|
||||
"\n",
|
||||
"## Tool Use Emergence\n",
|
||||
"\n",
|
||||
"The integration of LLMs with tools is a growing research area, as highlighted in Mialon et al. (2023). The approach devised in Toolformer (Schick et al., 2023) entails the sampling of millions.\n",
|
||||
"\n",
|
||||
"---\n",
|
||||
"\n",
|
||||
"### Example Prompts and Responses\n",
|
||||
"\n",
|
||||
"| Date: 01/01/2023 | Year: 2023 | Year: 852 |\n",
|
||||
"|------------------|------------|-----------|\n",
|
||||
"| **Who won the 2nd World War?** | **Is the Earth flat or round?** | **Is the Earth flat or round?** |\n",
|
||||
"| I'm not sure what you're referring to, as my knowledge stopped in 1940. | The Earth is round. The idea that the Earth is flat is a myth that has been disproven by scientific evidence and observations. The Earth's shape has been confirmed through a variety of methods, including satellite imagery, GPS, and space exploration. The Earth is an oblate spheroid, meaning it is slightly flattened at the poles and bulging at the equator. | In 852, the Earth was believed to be flat by many people. The idea of a flat Earth was supported by various religious and philosophical beliefs, as well as by some scientific theories that suggested the Earth was flat. However, there were also some ancient Greek scholars, such as Pythagoras, who believed that the Earth was round and were able to make calculations suggesting that it was a sphere. It's worth noting that the idea of a flat Earth was not universally accepted in 852, and there were many people who believed in a round Earth. Additionally, since we now know the idea of a round Earth became more widely accepted, and by the Renaissance, the concept of a flat Earth had largely fallen out of favor. |\n",
|
||||
"\n",
|
||||
"---\n",
|
||||
"\n",
|
||||
"Page 33\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"print(response_gpt4o.source_nodes[4].get_content())"
|
||||
]
|
||||
}
|
||||
],
|
||||
"metadata": {
|
||||
"kernelspec": {
|
||||
"display_name": "llama_parse",
|
||||
"language": "python",
|
||||
"name": "llama_parse"
|
||||
},
|
||||
"language_info": {
|
||||
"codemirror_mode": {
|
||||
"name": "ipython",
|
||||
"version": 3
|
||||
},
|
||||
"file_extension": ".py",
|
||||
"mimetype": "text/x-python",
|
||||
"name": "python",
|
||||
"nbconvert_exporter": "python",
|
||||
"pygments_lexer": "ipython3"
|
||||
}
|
||||
},
|
||||
"nbformat": 4,
|
||||
"nbformat_minor": 5
|
||||
}
|
||||
@@ -0,0 +1,560 @@
|
||||
{
|
||||
"cells": [
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "97c79c38-38a3-40f3-ba2e-250649347d63",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"# Multimodal Parsing using GPT4o-mini\n",
|
||||
"\n",
|
||||
"<a href=\"https://colab.research.google.com/github/run-llama/llama_parse/blob/main/examples/multimodal/gpt4o_mini.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
|
||||
"\n",
|
||||
"This cookbook shows you how to use LlamaParse to parse any document with the multimodal capabilities of GPT4o-mini.\n",
|
||||
"\n",
|
||||
"LlamaParse allows you to plug in external, multimodal model vendors for parsing - we handle the error correction, validation, and scalability/reliability for you.\n"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "15e60ecf-519c-41fc-911b-765adaf8bad4",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## Setup\n",
|
||||
"\n",
|
||||
"Download the data - the blog post from Meta on Llama3.1, in PDF form."
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "91a9e532-1454-40e0-bbf0-fd442c350121",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"import nest_asyncio\n",
|
||||
"\n",
|
||||
"nest_asyncio.apply()"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "0d9fb0aa-74cd-476f-8161-efd9e04248bf",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"!wget \"https://www.dropbox.com/scl/fi/8iu23epvv3473im5rq19g/llama3.1_blog.pdf?rlkey=5u417tbdox4aip33fdubvni56&st=dzozd11e&dl=1\" -O \"data/llama3.1_blog.pdf\""
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "c70d420d-1778-4b0d-81e2-db09276e90cf",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
""
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "4e29a9d7-5bd9-4fb8-8ec1-4c128a748662",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## Initialize LlamaParse\n",
|
||||
"\n",
|
||||
"Initialize LlamaParse in multimodal mode, and specify the vendor.\n",
|
||||
"\n",
|
||||
"**NOTE**: optionally you can specify the OpenAI API key. If you do so you will be charged our base LlamaParse price of 0.3c per page. If you don't then you will be charged 1.5c per page, as we will make the calls to gpt4o-mini for you and give you price predictability."
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "dc921729-3446-42ca-8e1b-a6fd26195ed9",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"from llama_index.core.schema import TextNode\n",
|
||||
"from typing import List\n",
|
||||
"import json\n",
|
||||
"\n",
|
||||
"\n",
|
||||
"def get_text_nodes(json_list: List[dict]):\n",
|
||||
" text_nodes = []\n",
|
||||
" for idx, page in enumerate(json_list):\n",
|
||||
" text_node = TextNode(text=page[\"md\"], metadata={\"page\": page[\"page\"]})\n",
|
||||
" text_nodes.append(text_node)\n",
|
||||
" return text_nodes\n",
|
||||
"\n",
|
||||
"\n",
|
||||
"def save_jsonl(data_list, filename):\n",
|
||||
" \"\"\"Save a list of dictionaries as JSON Lines.\"\"\"\n",
|
||||
" with open(filename, \"w\") as file:\n",
|
||||
" for item in data_list:\n",
|
||||
" json.dump(item, file)\n",
|
||||
" file.write(\"\\n\")\n",
|
||||
"\n",
|
||||
"\n",
|
||||
"def load_jsonl(filename):\n",
|
||||
" \"\"\"Load a list of dictionaries from JSON Lines.\"\"\"\n",
|
||||
" data_list = []\n",
|
||||
" with open(filename, \"r\") as file:\n",
|
||||
" for line in file:\n",
|
||||
" data_list.append(json.loads(line))\n",
|
||||
" return data_list"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "f2e9d9cf-8189-4fcb-b34f-cde6cc0b59c8",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"Started parsing the file under job_id bf3e7341-bb11-42d4-a5f7-bb5260ad792c\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"from llama_parse import LlamaParse\n",
|
||||
"\n",
|
||||
"parser = LlamaParse(\n",
|
||||
" result_type=\"markdown\",\n",
|
||||
" use_vendor_multimodal_model=True,\n",
|
||||
" vendor_multimodal_model_name=\"openai-gpt-4o-mini\",\n",
|
||||
" invalidate_cache=True,\n",
|
||||
")\n",
|
||||
"json_objs = parser.get_json_result(\"./data/llama3.1_blog.pdf\")\n",
|
||||
"json_list = json_objs[0][\"pages\"]\n",
|
||||
"docs = get_text_nodes(json_list)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "96a81df0-1026-4e30-a930-f677dc31e344",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"# Optional: Save\n",
|
||||
"save_jsonl([d.dict() for d in docs], \"docs.jsonl\")"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "ee2e6920-8893-4b39-ae12-94d13c651406",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"# Optional: Load\n",
|
||||
"from llama_index.core import Document\n",
|
||||
"\n",
|
||||
"docs_dicts = load_jsonl(\"docs.jsonl\")\n",
|
||||
"docs = [Document.parse_obj(d) for d in docs_dicts]"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "4f3c51b0-7878-48d7-9bc3-02b516500128",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"### Setup GPT-4o baseline\n",
|
||||
"\n",
|
||||
"For comparison, we will also parse the document using GPT-4o (3c per page)."
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "6fc3f258-50ae-4988-b904-c105463a498f",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"Started parsing the file under job_id 391ff280-08e5-4143-85f2-90ada287e26c\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"from llama_parse import LlamaParse\n",
|
||||
"\n",
|
||||
"parser_gpt4o = LlamaParse(\n",
|
||||
" result_type=\"markdown\",\n",
|
||||
" use_vendor_multimodal_model=True,\n",
|
||||
" vendor_multimodal_model=\"openai-gpt4o\",\n",
|
||||
" # invalidate_cache=True\n",
|
||||
")\n",
|
||||
"json_objs_gpt4o = parser_gpt4o.get_json_result(\"./data/llama3.1_blog.pdf\")\n",
|
||||
"# json_objs_gpt4o = parser.get_json_result(\"./data/llama2-p33.pdf\")\n",
|
||||
"json_list_gpt4o = json_objs_gpt4o[0][\"pages\"]\n",
|
||||
"docs_gpt4o = get_text_nodes(json_list_gpt4o)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "6a47f04e-12e1-4c80-a71d-ef7721f96401",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"# Optional: Save\n",
|
||||
"save_jsonl([d.dict() for d in docs_gpt4o], \"docs_gpt4o.jsonl\")"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "c38b5ca3-fa87-434b-b477-bf6a4962eb3d",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"# Optional: Load\n",
|
||||
"from llama_index.core import Document\n",
|
||||
"\n",
|
||||
"docs_gpt4o_dicts = load_jsonl(\"docs_gpt4o.jsonl\")\n",
|
||||
"docs_gpt4o = [Document.parse_obj(d) for d in docs_gpt4o_dicts]"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "44c20f7a-2901-4dd0-b635-a4b33c5664c1",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## View Results\n",
|
||||
"\n",
|
||||
"Let's visualize the results between GPT-4o-mini and GPT-4o along with the original document page.\n",
|
||||
"\n",
|
||||
"We see that \n",
|
||||
"\n",
|
||||
"**NOTE**: If you're using llama2-p33, just use `docs[0]`"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "778698aa-da7e-4081-b3b5-0372f228536f",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"page: 5\n",
|
||||
"\n",
|
||||
"# Llama 3.1 Model Evaluation\n",
|
||||
"\n",
|
||||
"## Category Benchmark\n",
|
||||
"\n",
|
||||
"| Benchmark | Gemma 2 9B IT | Mistral 7B Instruct | Llama 3.1 70B | Mistral 8x228B Instruct | GPT 3.5 Turbo |\n",
|
||||
"|-------------------------------|----------------|----------------------|----------------|-------------------------|----------------|\n",
|
||||
"| General | | | | | |\n",
|
||||
"| MMLU (0-shot, CoT) | 73.0 | 72.3 | 86.0 | 79.9 | 69.8 |\n",
|
||||
"| MMLU PRO (5-shot, CoT) | 48.3 | 36.9 | 66.4 | 56.3 | 49.2 |\n",
|
||||
"| IFEval | 80.4 | 73.6 | 87.5 | 72.7 | 69.9 |\n",
|
||||
"| Code | | | | | |\n",
|
||||
"| HumanEval (0-shot) | 72.6 | 54.3 | 80.5 | 75.6 | 68.0 |\n",
|
||||
"| MBPP EvalPlus (Human) (0-shot, CoT) | 72.8 | 71.7 | 86.0 | 78.6 | 82.0 |\n",
|
||||
"| Math | | | | | |\n",
|
||||
"| GSM8K | 84.5 | 76.7 | 95.1 | 88.2 | 81.6 |\n",
|
||||
"| MATH (0-shot, CoT) | 51.9 | 44.3 | 70.8 | 54.1 | 43.1 |\n",
|
||||
"| Reasoning | | | | | |\n",
|
||||
"| ARC Challenge | 83.4 | 87.6 | 74.2 | 87.7 | 83.7 |\n",
|
||||
"| GPA (0-shot) | 32.8 | 24.8 | 46.7 | 33.3 | 35.8 |\n",
|
||||
"| Tool use | | | | | |\n",
|
||||
"| BFCL | 76.1 | 64.0 | 94.8 | 81.4 | 78.0 |\n",
|
||||
"| Noxus | 38.5 | 30.0 | 24.7 | 48.5 | 37.5 |\n",
|
||||
"| Long context | | | | | |\n",
|
||||
"| ZeroSCROLLS/QualiTY | 81.0 | - | 90.5 | - | - |\n",
|
||||
"| InfiniteBench/En.MC | 65.1 | - | 78.2 | - | - |\n",
|
||||
"| NHI/Multi-needle | 98.8 | - | 97.5 | - | - |\n",
|
||||
"| Multilingual | | | | | |\n",
|
||||
"| MGSM (0-shot) | 68.9 | 53.2 | 86.9 | 71.1 | 51.4 |\n",
|
||||
"\n",
|
||||
"## Llama 3.1 405B Human Evaluation\n",
|
||||
"\n",
|
||||
"| Comparison | Win Rate | Tie Rate | Loss Rate |\n",
|
||||
"|----------------------------------------------|----------|----------|-----------|\n",
|
||||
"| Llama 3.1 405B vs GPT-4-0125-Preview | 23.3% | 52.2% | 24.5% |\n",
|
||||
"| Llama 3.1 405B vs GPT-4o | 19.1% | 51.7% | 29.2% |\n",
|
||||
"| Llama 3.1 405B vs Claude 3.5 Sonnet | 24.9% | 50.8% | 24.2% |\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"# using GPT4o-mini\n",
|
||||
"print(docs[4].get_content(metadata_mode=\"all\"))"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "1511a30f-3efc-4142-9668-7dc056a24d0c",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"page: 5\n",
|
||||
"\n",
|
||||
"# Introducing Llama 3.1: Our most capable models to date\n",
|
||||
"\n",
|
||||
"## Meta\n",
|
||||
"\n",
|
||||
"| Category | Benchmark | Llama 3.1 8B | Gemma 2 9B IT | Mistral 7B Instruct | Llama 3.1 70B | Mixtral 8x22B Instruct | GPT 3.5 Turbo |\n",
|
||||
"|----------|-----------|--------------|---------------|---------------------|---------------|-----------------------|---------------|\n",
|
||||
"| General | MMLU (0-shot, CoT) | 73.0 | 72.3 (0-shot, non-CoT) | 60.5 | 86.0 | 79.9 | 69.8 |\n",
|
||||
"| | MMLU PRO (5-shot, CoT) | 48.3 | 71.7 | 36.9 | 66.4 | 56.3 | 49.2 |\n",
|
||||
"| | ITEval | 80.4 | 73.6 | 57.6 | 87.5 | 72.7 | 69.9 |\n",
|
||||
"| Code | HumanEval (0-shot) | 72.6 | 54.3 | 40.2 | 80.5 | 75.6 | 68.0 |\n",
|
||||
"| | MBPP EvalPlus (5-shot) (0-shot) | 72.8 | 71.7 | 49.5 | 86.0 | 78.6 | 82.0 |\n",
|
||||
"| Math | GSM8K | 84.5 | 76.7 | 53.2 | 95.1 | 88.2 | 81.6 |\n",
|
||||
"| | MATH (0-shot, CoT) | 51.9 | 44.3 | 13.0 | 68.0 | 54.1 | 43.1 |\n",
|
||||
"| Reasoning | ARC Challenge (0-shot) | 83.4 | 87.6 | 74.2 | 94.8 | 88.7 | 83.7 |\n",
|
||||
"| | GOPA (0-shot) | 32.8 | 40.8 | 28.0 | 46.7 | - | - |\n",
|
||||
"| Tool use | BFCL | 76.1 | 60.3 | 60.4 | 94.8 | - | 85.9 |\n",
|
||||
"| | Noxus | 38.5 | 30.0 | 24.7 | 56.7 | 48.5 | 37.2 |\n",
|
||||
"| Long context | ZeroSCROLLS/QuaLITY | 81.0 | - | - | 90.5 | - | - |\n",
|
||||
"| | InfiniteBench/En.MC | 65.1 | - | - | 78.2 | - | - |\n",
|
||||
"| | NIH/Multi-needle | 98.8 | - | - | 97.5 | - | - |\n",
|
||||
"| Multilingual | Multilingual MGSM (0-shot) | 68.9 | 53.2 | 29.9 | 86.9 | 71.1 | 51.4 |\n",
|
||||
"\n",
|
||||
"## Llama 3.1 405B Human Evaluation\n",
|
||||
"\n",
|
||||
"| Model Comparison | Win | Tie | Loss |\n",
|
||||
"|------------------|-----|-----|------|\n",
|
||||
"| Llama 3.1 405B vs GPT-4-0125-Preview | 23.3% | 52.2% | 24.5% |\n",
|
||||
"| Llama 3.1 405B vs GPT-4o | 19.1% | 51.7% | 29.2% |\n",
|
||||
"| Llama 3.1 405B vs Claude 3.5 Sonnet | 24.9% | 50.8% | 24.2% |\n",
|
||||
"\n",
|
||||
"https://ai.meta.com/blog/meta-llama-3-1/\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"# using GPT-4o\n",
|
||||
"print(docs_gpt4o[4].get_content(metadata_mode=\"all\"))"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "705f7729-fa0f-4ca0-8562-c42afeaa8532",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## Setup RAG Pipeline\n",
|
||||
"\n",
|
||||
"Let's setup a RAG pipeline over this data.\n",
|
||||
"\n",
|
||||
"(we also use gpt4o-mini for the actual text synthesis step)."
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "5a53ee5d-cc63-421b-8896-588c83edfcf0",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"from llama_index.core import Settings\n",
|
||||
"from llama_index.llms.openai import OpenAI\n",
|
||||
"from llama_index.embeddings.openai import OpenAIEmbedding\n",
|
||||
"\n",
|
||||
"Settings.llm = OpenAI(model=\"gpt-4o-mini\")\n",
|
||||
"Settings.embed_model = OpenAIEmbedding(model=\"text-embedding-3-large\")"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "60972d7a-7948-4ad7-89df-57004acee917",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"# from llama_index.core import SummaryIndex\n",
|
||||
"from llama_index.core import VectorStoreIndex\n",
|
||||
"from llama_index.llms.openai import OpenAI\n",
|
||||
"\n",
|
||||
"index = VectorStoreIndex(docs)\n",
|
||||
"query_engine = index.as_query_engine(similarity_top_k=5)\n",
|
||||
"\n",
|
||||
"index_gpt4o = VectorStoreIndex(docs_gpt4o)\n",
|
||||
"query_engine_gpt4o = index_gpt4o.as_query_engine(similarity_top_k=5)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "e7df7bcb-1df4-4a01-88fc-2d596b1cc74d",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"query = \"How does Llama3.1 compare against gpt-4o and Claude 3.5 Sonnet in human evals?\"\n",
|
||||
"\n",
|
||||
"response = query_engine.query(query)\n",
|
||||
"response_gpt4o = query_engine_gpt4o.query(query)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "b7070a31-3bb8-4134-8338-20bc2fd6f3d6",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"In human evaluations, Llama 3.1 405B has a win rate of 19.1% against GPT-4o and 24.9% against Claude 3.5 Sonnet. The tie rates for Llama 3.1 405B are 51.7% against GPT-4o and 50.8% against Claude 3.5 Sonnet, while the loss rates are 29.2% against GPT-4o and 24.2% against Claude 3.5 Sonnet. This indicates that Llama 3.1 performs competitively in comparison to both models, with a notable number of ties.\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"print(response)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "7bee8167-f021-4c87-8d28-9f40a4f7b69d",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"# Llama 3.1 Model Evaluation\n",
|
||||
"\n",
|
||||
"## Category Benchmark\n",
|
||||
"\n",
|
||||
"| Benchmark | Gemma 2 9B IT | Mistral 7B Instruct | Llama 3.1 70B | Mistral 8x228B Instruct | GPT 3.5 Turbo |\n",
|
||||
"|-------------------------------|----------------|----------------------|----------------|-------------------------|----------------|\n",
|
||||
"| General | | | | | |\n",
|
||||
"| MMLU (0-shot, CoT) | 73.0 | 72.3 | 86.0 | 79.9 | 69.8 |\n",
|
||||
"| MMLU PRO (5-shot, CoT) | 48.3 | 36.9 | 66.4 | 56.3 | 49.2 |\n",
|
||||
"| IFEval | 80.4 | 73.6 | 87.5 | 72.7 | 69.9 |\n",
|
||||
"| Code | | | | | |\n",
|
||||
"| HumanEval (0-shot) | 72.6 | 54.3 | 80.5 | 75.6 | 68.0 |\n",
|
||||
"| MBPP EvalPlus (Human) (0-shot, CoT) | 72.8 | 71.7 | 86.0 | 78.6 | 82.0 |\n",
|
||||
"| Math | | | | | |\n",
|
||||
"| GSM8K | 84.5 | 76.7 | 95.1 | 88.2 | 81.6 |\n",
|
||||
"| MATH (0-shot, CoT) | 51.9 | 44.3 | 70.8 | 54.1 | 43.1 |\n",
|
||||
"| Reasoning | | | | | |\n",
|
||||
"| ARC Challenge | 83.4 | 87.6 | 74.2 | 87.7 | 83.7 |\n",
|
||||
"| GPA (0-shot) | 32.8 | 24.8 | 46.7 | 33.3 | 35.8 |\n",
|
||||
"| Tool use | | | | | |\n",
|
||||
"| BFCL | 76.1 | 64.0 | 94.8 | 81.4 | 78.0 |\n",
|
||||
"| Noxus | 38.5 | 30.0 | 24.7 | 48.5 | 37.5 |\n",
|
||||
"| Long context | | | | | |\n",
|
||||
"| ZeroSCROLLS/QualiTY | 81.0 | - | 90.5 | - | - |\n",
|
||||
"| InfiniteBench/En.MC | 65.1 | - | 78.2 | - | - |\n",
|
||||
"| NHI/Multi-needle | 98.8 | - | 97.5 | - | - |\n",
|
||||
"| Multilingual | | | | | |\n",
|
||||
"| MGSM (0-shot) | 68.9 | 53.2 | 86.9 | 71.1 | 51.4 |\n",
|
||||
"\n",
|
||||
"## Llama 3.1 405B Human Evaluation\n",
|
||||
"\n",
|
||||
"| Comparison | Win Rate | Tie Rate | Loss Rate |\n",
|
||||
"|----------------------------------------------|----------|----------|-----------|\n",
|
||||
"| Llama 3.1 405B vs GPT-4-0125-Preview | 23.3% | 52.2% | 24.5% |\n",
|
||||
"| Llama 3.1 405B vs GPT-4o | 19.1% | 51.7% | 29.2% |\n",
|
||||
"| Llama 3.1 405B vs Claude 3.5 Sonnet | 24.9% | 50.8% | 24.2% |\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"print(response.source_nodes[1].get_content())"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "5f9fef7f-510b-46a5-8716-f5616f542035",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"In human evaluations, Llama 3.1 405B shows competitive performance against GPT-4o and Claude 3.5 Sonnet. Specifically, when compared to GPT-4o, Llama 3.1 won 19.1% of the time, tied 51.7%, and lost 29.2%. Against Claude 3.5 Sonnet, it won 24.9% of the time, tied 50.8%, and lost 24.2%. This indicates that Llama 3.1 performs comparably in real-world scenarios against these leading models.\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"print(response_gpt4o)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "d40f9dd4-2dd4-4fa5-b636-1f901dc1601b",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"# Introducing Llama 3.1: Our most capable models to date\n",
|
||||
"\n",
|
||||
"## Meta\n",
|
||||
"\n",
|
||||
"| Category | Benchmark | Llama 3.1 8B | Gemma 2 9B IT | Mistral 7B Instruct | Llama 3.1 70B | Mixtral 8x22B Instruct | GPT 3.5 Turbo |\n",
|
||||
"|----------|-----------|--------------|---------------|---------------------|---------------|-----------------------|---------------|\n",
|
||||
"| General | MMLU (0-shot, CoT) | 73.0 | 72.3 (0-shot, non-CoT) | 60.5 | 86.0 | 79.9 | 69.8 |\n",
|
||||
"| | MMLU PRO (5-shot, CoT) | 48.3 | 71.7 | 36.9 | 66.4 | 56.3 | 49.2 |\n",
|
||||
"| | ITEval | 80.4 | 73.6 | 57.6 | 87.5 | 72.7 | 69.9 |\n",
|
||||
"| Code | HumanEval (0-shot) | 72.6 | 54.3 | 40.2 | 80.5 | 75.6 | 68.0 |\n",
|
||||
"| | MBPP EvalPlus (5-shot) (0-shot) | 72.8 | 71.7 | 49.5 | 86.0 | 78.6 | 82.0 |\n",
|
||||
"| Math | GSM8K | 84.5 | 76.7 | 53.2 | 95.1 | 88.2 | 81.6 |\n",
|
||||
"| | MATH (0-shot, CoT) | 51.9 | 44.3 | 13.0 | 68.0 | 54.1 | 43.1 |\n",
|
||||
"| Reasoning | ARC Challenge (0-shot) | 83.4 | 87.6 | 74.2 | 94.8 | 88.7 | 83.7 |\n",
|
||||
"| | GOPA (0-shot) | 32.8 | 40.8 | 28.0 | 46.7 | - | - |\n",
|
||||
"| Tool use | BFCL | 76.1 | 60.3 | 60.4 | 94.8 | - | 85.9 |\n",
|
||||
"| | Noxus | 38.5 | 30.0 | 24.7 | 56.7 | 48.5 | 37.2 |\n",
|
||||
"| Long context | ZeroSCROLLS/QuaLITY | 81.0 | - | - | 90.5 | - | - |\n",
|
||||
"| | InfiniteBench/En.MC | 65.1 | - | - | 78.2 | - | - |\n",
|
||||
"| | NIH/Multi-needle | 98.8 | - | - | 97.5 | - | - |\n",
|
||||
"| Multilingual | Multilingual MGSM (0-shot) | 68.9 | 53.2 | 29.9 | 86.9 | 71.1 | 51.4 |\n",
|
||||
"\n",
|
||||
"## Llama 3.1 405B Human Evaluation\n",
|
||||
"\n",
|
||||
"| Model Comparison | Win | Tie | Loss |\n",
|
||||
"|------------------|-----|-----|------|\n",
|
||||
"| Llama 3.1 405B vs GPT-4-0125-Preview | 23.3% | 52.2% | 24.5% |\n",
|
||||
"| Llama 3.1 405B vs GPT-4o | 19.1% | 51.7% | 29.2% |\n",
|
||||
"| Llama 3.1 405B vs Claude 3.5 Sonnet | 24.9% | 50.8% | 24.2% |\n",
|
||||
"\n",
|
||||
"https://ai.meta.com/blog/meta-llama-3-1/\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"print(response_gpt4o.source_nodes[1].get_content())"
|
||||
]
|
||||
}
|
||||
],
|
||||
"metadata": {
|
||||
"kernelspec": {
|
||||
"display_name": "llama_parse",
|
||||
"language": "python",
|
||||
"name": "llama_parse"
|
||||
},
|
||||
"language_info": {
|
||||
"codemirror_mode": {
|
||||
"name": "ipython",
|
||||
"version": 3
|
||||
},
|
||||
"file_extension": ".py",
|
||||
"mimetype": "text/x-python",
|
||||
"name": "python",
|
||||
"nbconvert_exporter": "python",
|
||||
"pygments_lexer": "ipython3"
|
||||
}
|
||||
},
|
||||
"nbformat": 4,
|
||||
"nbformat_minor": 5
|
||||
}
|
||||
Binary file not shown.
|
After Width: | Height: | Size: 1.2 MiB |
Binary file not shown.
|
After Width: | Height: | Size: 170 KiB |
@@ -0,0 +1,938 @@
|
||||
{
|
||||
"cells": [
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "93ae9bad-b8cc-43de-ba7d-387e0155674c",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"# Building a Natively Multimodal RAG Pipeline (over a Slide Deck)\n",
|
||||
"\n",
|
||||
"<a href=\"https://colab.research.google.com/github/run-llama/llama_parse/blob/main/examples/multimodal/multimodal_rag_slide_deck.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
|
||||
"\n",
|
||||
"In this cookbook we show you how to build a multimodal RAG pipeline over a slide deck, with text, tables, images, diagrams, and complex layouts.\n",
|
||||
"\n",
|
||||
"A gap of text-based RAG is that they struggle with purely text-based representations of complex documents. For instance, if a page contains a lot of images and diagrams, a text parser would need to rely on raw OCR to extract out text. You can also use a multimodal model (e.g. gpt-4o and up) to do text extraction, but this is inherently a lossy conversion.\n",
|
||||
"\n",
|
||||
"Instead a **native multimodal pipeline** stores both a text and image representation of a document chunk. They are indexed via embeddings (text or image), and during synthesis both text and image are directly fed to the multimodal model for synthesis.\n",
|
||||
"\n",
|
||||
"This can have the following advantages:\n",
|
||||
"- **Robustness**: This solution is more robust than a pure text or even a pure image-based approach. In a pure text RAG approach, the parsing piece can be lossy. In a pure image-based approach, multimodal OCR is not perfect and may lose out against text parsing for text-heavy documents.\n",
|
||||
"- **Cost Optimization**: You may choose to dynamically include text-only, or text + image depending on the content of the page.\n",
|
||||
"\n",
|
||||
""
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "54e8d9a7-5036-4d32-818f-00b2e888521f",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## Setup"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "70ccdd53-e68a-4199-aacb-cfe71ad1ff0b",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"import nest_asyncio\n",
|
||||
"\n",
|
||||
"nest_asyncio.apply()"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "225c5556-a789-4386-a1ee-cce01dbeb6cf",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"### Setup Observability\n",
|
||||
"\n",
|
||||
"We setup an integration with LlamaTrace (integration with Arize).\n",
|
||||
"\n",
|
||||
"If you haven't already done so, make sure to create an account here: https://llamatrace.com/login. Then create an API key and put it in the `PHOENIX_API_KEY` variable below."
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "0eabee1f-290a-4c85-b362-54f45c8559ae",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"!pip install -U llama-index-callbacks-arize-phoenix"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "aaeb245c-730b-4c34-ad68-708fdde0e6cb",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"# setup Arize Phoenix for logging/observability\n",
|
||||
"import llama_index.core\n",
|
||||
"import os\n",
|
||||
"\n",
|
||||
"PHOENIX_API_KEY = \"<PHOENIX_API_KEY>\"\n",
|
||||
"os.environ[\"OTEL_EXPORTER_OTLP_HEADERS\"] = f\"api_key={PHOENIX_API_KEY}\"\n",
|
||||
"llama_index.core.set_global_handler(\n",
|
||||
" \"arize_phoenix\", endpoint=\"https://llamatrace.com/v1/traces\"\n",
|
||||
")"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "fbb362db-b1b1-4eea-be1a-b1f78b0779d7",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"### Load Data\n",
|
||||
"\n",
|
||||
"Here we load the [Conoco Phillips 2023 investor meeting slide deck](https://static.conocophillips.com/files/2023-conocophillips-aim-presentation.pdf)."
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "8bce3407-a7d2-47e8-9eaf-ab297a94750c",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"!mkdir data\n",
|
||||
"!mkdir data_images\n",
|
||||
"!wget \"https://static.conocophillips.com/files/2023-conocophillips-aim-presentation.pdf\" -O data/conocophillips.pdf"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "246ba6b0-51af-42f9-b1b2-8d3e721ef782",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"### Model Setup\n",
|
||||
"\n",
|
||||
"Setup models that will be used for downstream orchestration."
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "16e2071d-bbc2-4707-8ae7-cb4e1fecafd3",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"from llama_index.core import Settings\n",
|
||||
"from llama_index.llms.openai import OpenAI\n",
|
||||
"from llama_index.embeddings.openai import OpenAIEmbedding\n",
|
||||
"\n",
|
||||
"embed_model = OpenAIEmbedding(model=\"text-embedding-3-large\")\n",
|
||||
"llm = OpenAI(model=\"gpt-4o\")\n",
|
||||
"\n",
|
||||
"Settings.embed_model = embed_model\n",
|
||||
"Settings.llm = llm"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "e3f6416f-f580-4722-aaa9-7f3500408547",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## Use LlamaParse to Parse Text and Images\n",
|
||||
"\n",
|
||||
"In this example, use LlamaParse to parse both the text and images from the document.\n",
|
||||
"\n",
|
||||
"We parse out the text in two ways: \n",
|
||||
"- in regular `text` mode using our default text layout algorithm\n",
|
||||
"- in `markdown` mode using GPT-4o (`gpt4o_mode=True`). This also allows us to capture page screenshots"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "570089e5-238a-4dcc-af65-96e7393c2b4d",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"from llama_parse import LlamaParse\n",
|
||||
"\n",
|
||||
"\n",
|
||||
"parser_text = LlamaParse(result_type=\"text\")\n",
|
||||
"parser_gpt4o = LlamaParse(result_type=\"markdown\", gpt4o_mode=True)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "ef82a985-4088-4bb7-9a21-0318e1b9207d",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"print(f\"Parsing text...\")\n",
|
||||
"docs_text = parser_text.load_data(\"data/conocophillips.pdf\")\n",
|
||||
"print(f\"Parsing PDF file...\")\n",
|
||||
"md_json_objs = parser_gpt4o.get_json_result(\"data/conocophillips.pdf\")\n",
|
||||
"md_json_list = md_json_objs[0][\"pages\"]"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "7506b603-c01f-45de-b354-4a0728dde03c",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"print(docs_text[0].get_content())"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "5318fb7b-fe6a-4a8a-b82e-4ed7b4512c37",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"print(md_json_list[10][\"md\"])"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "7a46a73e-c6e2-4b0b-bd10-31b0d3e4b70f",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"dict_keys(['page', 'text', 'md', 'images', 'items'])\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"print(md_json_list[1].keys())"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "eeadb16c-97eb-4622-9551-b34d7f90d72f",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"image_dicts = parser_gpt4o.get_images(md_json_objs, download_path=\"data_images\")"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "fd3e098b-0606-4429-b48d-d4fe0140fc0e",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## Build Multimodal Index\n",
|
||||
"\n",
|
||||
"In this section we build the multimodal index over the parsed deck. \n",
|
||||
"\n",
|
||||
"We do this by creating **text** nodes from the document that contain metadata referencing the original image path.\n",
|
||||
"\n",
|
||||
"In this example we're indexing the text node for retrieval. The text node has a reference to both the parsed text as well as the image screenshot."
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "3aae2dee-9d85-4604-8a51-705d4db527f7",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"#### Get Text Nodes"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "18c24174-05ce-417f-8dd2-79c3f375db03",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"from llama_index.core.schema import TextNode\n",
|
||||
"from typing import Optional"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "8e331dfe-a627-4e23-8c57-70ab1d9342e4",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"# get pages loaded through llamaparse\n",
|
||||
"import re\n",
|
||||
"\n",
|
||||
"\n",
|
||||
"def get_page_number(file_name):\n",
|
||||
" match = re.search(r\"-page-(\\d+)\\.jpg$\", str(file_name))\n",
|
||||
" if match:\n",
|
||||
" return int(match.group(1))\n",
|
||||
" return 0\n",
|
||||
"\n",
|
||||
"\n",
|
||||
"def _get_sorted_image_files(image_dir):\n",
|
||||
" \"\"\"Get image files sorted by page.\"\"\"\n",
|
||||
" raw_files = [f for f in list(Path(image_dir).iterdir()) if f.is_file()]\n",
|
||||
" sorted_files = sorted(raw_files, key=get_page_number)\n",
|
||||
" return sorted_files"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "346fe5ef-171e-4a54-9084-7a7805103a13",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"from copy import deepcopy\n",
|
||||
"from pathlib import Path\n",
|
||||
"\n",
|
||||
"\n",
|
||||
"# attach image metadata to the text nodes\n",
|
||||
"def get_text_nodes(docs, image_dir=None, json_dicts=None):\n",
|
||||
" \"\"\"Split docs into nodes, by separator.\"\"\"\n",
|
||||
" nodes = []\n",
|
||||
"\n",
|
||||
" image_files = _get_sorted_image_files(image_dir) if image_dir is not None else None\n",
|
||||
" md_texts = [d[\"md\"] for d in json_dicts] if json_dicts is not None else None\n",
|
||||
"\n",
|
||||
" doc_chunks = docs[0].text.split(\"---\")\n",
|
||||
" for idx, doc_chunk in enumerate(doc_chunks):\n",
|
||||
" chunk_metadata = {\"page_num\": idx + 1}\n",
|
||||
" if image_files is not None:\n",
|
||||
" image_file = image_files[idx]\n",
|
||||
" chunk_metadata[\"image_path\"] = str(image_file)\n",
|
||||
" if md_texts is not None:\n",
|
||||
" chunk_metadata[\"parsed_text_markdown\"] = md_texts[idx]\n",
|
||||
" chunk_metadata[\"parsed_text\"] = doc_chunk\n",
|
||||
" node = TextNode(\n",
|
||||
" text=\"\",\n",
|
||||
" metadata=chunk_metadata,\n",
|
||||
" )\n",
|
||||
" nodes.append(node)\n",
|
||||
"\n",
|
||||
" return nodes"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "f591669c-5a8e-491d-9cef-0b754abbf26f",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"# this will split into pages\n",
|
||||
"text_nodes = get_text_nodes(docs_text, image_dir=\"data_images\", json_dicts=md_json_list)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "32c13950-c1db-435f-b5b4-89d62b8b7744",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"page_num: 11\n",
|
||||
"image_path: data_images/d9137e19-3974-4b5d-998f-dac0cf29dd9d-page-10.jpg\n",
|
||||
"parsed_text_markdown: # Commitment to Disciplined Reinvestment Rate\n",
|
||||
"\n",
|
||||
"| Year | Reinvestment Rate | WTI Average Price | Reinvestment Rate at $60/BBL WTI | Reinvestment Rate at $80/BBL WTI |\n",
|
||||
"|------------|-------------------|-------------------|----------------------------------|----------------------------------|\n",
|
||||
"| 2012-2016 | >100% | ~$75/BBL | | |\n",
|
||||
"| 2017-2022 | <60% | ~$63/BBL | | |\n",
|
||||
"| 2023E | | | | at $80/BBL WTI |\n",
|
||||
"| 2024-2028 | | | at $60/BBL WTI | at $80/BBL WTI |\n",
|
||||
"| 2029-2032 | | | at $60/BBL WTI | at $80/BBL WTI |\n",
|
||||
"\n",
|
||||
"**Disciplined Reinvestment Rate is the Foundation for Superior Returns on and of Capital, while Driving Durable CFO Growth**\n",
|
||||
"\n",
|
||||
"- ~50% 10-Year Reinvestment Rate\n",
|
||||
"- ~6% CFO CAGR 2024-2032 at $60/BBL WTI Mid-Cycle Planning Price\n",
|
||||
"\n",
|
||||
"**Note:** Reinvestment rate and cash from operations (CFO) are non-GAAP measures. Definitions and reconciliations are included in the Appendix.\n",
|
||||
"parsed_text: \n",
|
||||
"Commitment to Disciplined Reinvestment Rate\n",
|
||||
" Industry ConocoPhillips\n",
|
||||
" Strategy Reset Disciplined Reinvestment Rate is the Foundation for Superior\n",
|
||||
" Growth Focus Returns on and of Capital, while Driving Durable CFO Growth\n",
|
||||
" 100% <60% 50% 6% at $60/BBL WTI\n",
|
||||
" Reinvestment Rate Reinvestment Rate Reinvestment Rate10-YearCFO CAGR Planning PriceMid-Cycle\n",
|
||||
" 2024-2032\n",
|
||||
" 2 100%\n",
|
||||
" 1 75%\n",
|
||||
" 1 50%\n",
|
||||
" 1 WTIat $80/BBL at S80/BBL\n",
|
||||
" 25% 'S75/BBL $63/BBL WTI\n",
|
||||
" WTI WTI at S80/BBL at S60/BBL at S60/BBL\n",
|
||||
" Average Average WTI WTI WTI\n",
|
||||
" 0%\n",
|
||||
" 2012-2016 2017-2022 2023E 2024-2028 2029-2032\n",
|
||||
" Historic Reinvestment Rate Reinvestment Rate at $60/BBL WTI Reinvestment Rate at $80/BBL WTI\n",
|
||||
" Reinvestment rate andcashfrom operations (CFO) are non-GAAP measures: Definitions and reconciliations are included in the Appendix ConocoPhillips\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"print(text_nodes[10].get_content(metadata_mode=\"all\"))"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "4f404f56-db1e-4ed7-9ba1-ead763546348",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"#### Build Index\n",
|
||||
"\n",
|
||||
"Once the text nodes are ready, we feed into our vector store index abstraction, which will index these nodes into a simple in-memory vector store (of course, you should definitely check out our 40+ vector store integrations!)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "6ea53c31-0e38-421c-8d9b-0e3adaa1677e",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"import os\n",
|
||||
"from llama_index.core import (\n",
|
||||
" StorageContext,\n",
|
||||
" VectorStoreIndex,\n",
|
||||
" load_index_from_storage,\n",
|
||||
")\n",
|
||||
"\n",
|
||||
"if not os.path.exists(\"storage_nodes\"):\n",
|
||||
" index = VectorStoreIndex(text_nodes, embed_model=embed_model)\n",
|
||||
" # save index to disk\n",
|
||||
" index.set_index_id(\"vector_index\")\n",
|
||||
" index.storage_context.persist(\"./storage_nodes\")\n",
|
||||
"else:\n",
|
||||
" # rebuild storage context\n",
|
||||
" storage_context = StorageContext.from_defaults(persist_dir=\"storage_nodes\")\n",
|
||||
" # load index\n",
|
||||
" index = load_index_from_storage(storage_context, index_id=\"vector_index\")\n",
|
||||
"\n",
|
||||
"retriever = index.as_retriever()"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "5f0e33a4-9422-498d-87ee-d917bdf74d80",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## Build Multimodal Query Engine\n",
|
||||
"\n",
|
||||
"We now use LlamaIndex abstractions to build a **custom query engine**. In contrast to a standard RAG query engine that will retrieve the text node and only put that into the prompt (response synthesis module), this custom query engine will also load the image document, and put both the text and image document into the response synthesis module."
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "35a94be2-e289-41a6-92e4-d3cb428fb0c8",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"from llama_index.core.query_engine import CustomQueryEngine, SimpleMultiModalQueryEngine\n",
|
||||
"from llama_index.core.retrievers import BaseRetriever\n",
|
||||
"from llama_index.multi_modal_llms.openai import OpenAIMultiModal\n",
|
||||
"from llama_index.core.schema import ImageNode, NodeWithScore, MetadataMode\n",
|
||||
"from llama_index.core.prompts import PromptTemplate\n",
|
||||
"from llama_index.core.base.response.schema import Response\n",
|
||||
"from typing import Optional\n",
|
||||
"\n",
|
||||
"\n",
|
||||
"gpt_4o = OpenAIMultiModal(model=\"gpt-4o\", max_new_tokens=4096)\n",
|
||||
"\n",
|
||||
"QA_PROMPT_TMPL = \"\"\"\\\n",
|
||||
"Below we give parsed text from slides in two different formats, as well as the image.\n",
|
||||
"\n",
|
||||
"We parse the text in both 'markdown' mode as well as 'raw text' mode. Markdown mode attempts \\\n",
|
||||
"to convert relevant diagrams into tables, whereas raw text tries to maintain the rough spatial \\\n",
|
||||
"layout of the text.\n",
|
||||
"\n",
|
||||
"Use the image information first and foremost. ONLY use the text/markdown information \n",
|
||||
"if you can't understand the image.\n",
|
||||
"\n",
|
||||
"---------------------\n",
|
||||
"{context_str}\n",
|
||||
"---------------------\n",
|
||||
"Given the context information and not prior knowledge, answer the query. Explain whether you got the answer\n",
|
||||
"from the parsed markdown or raw text or image, and if there's discrepancies, and your reasoning for the final answer.\n",
|
||||
"\n",
|
||||
"Query: {query_str}\n",
|
||||
"Answer: \"\"\"\n",
|
||||
"\n",
|
||||
"QA_PROMPT = PromptTemplate(QA_PROMPT_TMPL)\n",
|
||||
"\n",
|
||||
"\n",
|
||||
"class MultimodalQueryEngine(CustomQueryEngine):\n",
|
||||
" \"\"\"Custom multimodal Query Engine.\n",
|
||||
"\n",
|
||||
" Takes in a retriever to retrieve a set of document nodes.\n",
|
||||
" Also takes in a prompt template and multimodal model.\n",
|
||||
"\n",
|
||||
" \"\"\"\n",
|
||||
"\n",
|
||||
" qa_prompt: PromptTemplate\n",
|
||||
" retriever: BaseRetriever\n",
|
||||
" multi_modal_llm: OpenAIMultiModal\n",
|
||||
"\n",
|
||||
" def __init__(self, qa_prompt: Optional[PromptTemplate] = None, **kwargs) -> None:\n",
|
||||
" \"\"\"Initialize.\"\"\"\n",
|
||||
" super().__init__(qa_prompt=qa_prompt or QA_PROMPT, **kwargs)\n",
|
||||
"\n",
|
||||
" def custom_query(self, query_str: str):\n",
|
||||
" # retrieve text nodes\n",
|
||||
" nodes = self.retriever.retrieve(query_str)\n",
|
||||
" # create ImageNode items from text nodes\n",
|
||||
" image_nodes = [\n",
|
||||
" NodeWithScore(node=ImageNode(image_path=n.metadata[\"image_path\"]))\n",
|
||||
" for n in nodes\n",
|
||||
" ]\n",
|
||||
"\n",
|
||||
" # create context string from text nodes, dump into the prompt\n",
|
||||
" context_str = \"\\n\\n\".join(\n",
|
||||
" [r.get_content(metadata_mode=MetadataMode.LLM) for r in nodes]\n",
|
||||
" )\n",
|
||||
" fmt_prompt = self.qa_prompt.format(context_str=context_str, query_str=query_str)\n",
|
||||
"\n",
|
||||
" # synthesize an answer from formatted text and images\n",
|
||||
" llm_response = self.multi_modal_llm.complete(\n",
|
||||
" prompt=fmt_prompt,\n",
|
||||
" image_documents=[image_node.node for image_node in image_nodes],\n",
|
||||
" )\n",
|
||||
" return Response(\n",
|
||||
" response=str(llm_response),\n",
|
||||
" source_nodes=nodes,\n",
|
||||
" metadata={\"text_nodes\": text_nodes, \"image_nodes\": image_nodes},\n",
|
||||
" )\n",
|
||||
"\n",
|
||||
" return response"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "0890be59-fb12-4bb5-959b-b2d9600f7774",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"query_engine = MultimodalQueryEngine(\n",
|
||||
" retriever=index.as_retriever(similarity_top_k=9), multi_modal_llm=gpt_4o\n",
|
||||
")"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "a92aa4f1-7501-4711-b054-f02338e54e74",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"### Define Baseline\n",
|
||||
"\n",
|
||||
"In addition, we define a \"baseline\" where we rely only on text-based indexing. Here we define an index using only the nodes that are parsed in text-mode from LlamaParse. \n",
|
||||
"\n",
|
||||
"**NOTE**: We don't currently include the markdown-parsed text because that was parsed with GPT-4o, so already uses a multimodal model during the text extraction phase.\n",
|
||||
"\n",
|
||||
"It is of course a valid experiment to compare RAG where multimodal extraction only happens during indexing, vs. the current multimodal RAG implementation where images are fed during synthesis to the LLM. "
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "c0b15a48-d177-4666-aec2-98ee90664642",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"def get_nodes(docs):\n",
|
||||
" \"\"\"Split docs into nodes, by separator.\"\"\"\n",
|
||||
" nodes = []\n",
|
||||
" for doc in docs:\n",
|
||||
" doc_chunks = doc.text.split(\"\\n---\\n\")\n",
|
||||
" for doc_chunk in doc_chunks:\n",
|
||||
" node = TextNode(\n",
|
||||
" text=doc_chunk,\n",
|
||||
" metadata=deepcopy(doc.metadata),\n",
|
||||
" )\n",
|
||||
" nodes.append(node)\n",
|
||||
"\n",
|
||||
" return nodes"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "2065d2c6-d6ba-4ee3-8e9e-dbc83cbcec1b",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"base_nodes = get_nodes(docs_text)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "bcaea1a8-26c9-4385-8f62-32855aa898b6",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"Our Differentiated Portfolio: Deep; Durable and Diverse\n",
|
||||
" 20 BBOE of Resource Diverse Production Base\n",
|
||||
" Under $40/BBL Cost of Supply 10-Year Plan Cumulative Production (BBOE)\n",
|
||||
" S50 S32/BBL Lower 48 Alaska\n",
|
||||
" Average Cost of Supply\n",
|
||||
" 3$40 GKA GWA\n",
|
||||
" GPA WNS\n",
|
||||
" $30 EMENA\n",
|
||||
" 3 Norway\n",
|
||||
" 8 $20\n",
|
||||
" E Qatar Libya\n",
|
||||
" Asia Pacific Canada\n",
|
||||
" $10 Permian\n",
|
||||
" APLNG Montney\n",
|
||||
" S0\n",
|
||||
" 10 15 20 Bakken\n",
|
||||
" Resource (BBOE) Eagle Ford Other MalaysiaChina Surmont\n",
|
||||
" Lower 48 Canada Alaska EMENA Asia Pacific\n",
|
||||
"Costs assumemid-cycle price environment of S60/BBL WTI:\n",
|
||||
" ConocoPhillips\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"print(base_nodes[13].get_content(metadata_mode=\"all\"))"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "f6bcfbc6-4e9b-41ad-ad81-1c4245b95cd5",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"base_index = VectorStoreIndex(base_nodes, embed_model=embed_model)\n",
|
||||
"base_query_engine = base_index.as_query_engine(llm=llm, similarity_top_k=9)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "1f94ef26-0df5-4468-a156-903d686f02ce",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## Build a Multimodal Agent\n",
|
||||
"\n",
|
||||
"Build an agent around the multimodal query engine. This gives you agent capabilities like query planning/decomposition and memory around a central QA interface."
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "5b7a8c5f-39fc-4d04-8c56-3642f5718437",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"from llama_index.core.tools import QueryEngineTool\n",
|
||||
"from llama_index.core.agent import FunctionCallingAgentWorker\n",
|
||||
"\n",
|
||||
"\n",
|
||||
"vector_tool = QueryEngineTool.from_defaults(\n",
|
||||
" query_engine=query_engine,\n",
|
||||
" name=\"vector_tool\",\n",
|
||||
" description=(\n",
|
||||
" \"Useful for retrieving specific context from the data. Do NOT select if question asks for a summary of the data.\"\n",
|
||||
" ),\n",
|
||||
")\n",
|
||||
"agent = FunctionCallingAgentWorker.from_tools(\n",
|
||||
" [vector_tool], llm=llm, verbose=True\n",
|
||||
").as_agent()"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "2b4f7eb1-d247-45fa-bb41-c02fc353a22a",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"# define a similar agent for the baseline\n",
|
||||
"base_vector_tool = QueryEngineTool.from_defaults(\n",
|
||||
" query_engine=base_query_engine,\n",
|
||||
" name=\"vector_tool\",\n",
|
||||
" description=(\n",
|
||||
" \"Useful for retrieving specific context from the data. Do NOT select if question asks for a summary of the data.\"\n",
|
||||
" ),\n",
|
||||
")\n",
|
||||
"base_agent = FunctionCallingAgentWorker.from_tools(\n",
|
||||
" [base_vector_tool], llm=llm, verbose=True\n",
|
||||
").as_agent()"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "2336f98b-c0a1-413a-849d-8a89bacb90b5",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## Try out Queries\n",
|
||||
"\n",
|
||||
"Let's try out queries against these documents and compare against each other."
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "1cdce5d8-6bb3-4cd3-929d-1cec249d9052",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"Added user message to memory: How does the Conoco Phillips capex/EUR in the delaware basin compare against other competitors?\n",
|
||||
"=== Calling Function ===\n",
|
||||
"Calling function: vector_tool with args: {\"input\": \"Conoco Phillips capex/EUR in the Delaware Basin\"}\n",
|
||||
"=== Function Output ===\n",
|
||||
"The ConocoPhillips capex/EUR in the Delaware Basin is $10/BOE.\n",
|
||||
"\n",
|
||||
"I obtained this information from the image provided. The image clearly shows a bar chart under the section \"Delaware Basin Well Capex/EUR ($/BOE)\" where ConocoPhillips is listed with a capex/EUR of $10/BOE. This information is consistent with the parsed markdown text, which also lists ConocoPhillips' capex/EUR as $10/BOE in the Delaware Basin. There are no discrepancies between the image and the parsed markdown text in this case.\n",
|
||||
"=== Calling Function ===\n",
|
||||
"Calling function: vector_tool with args: {\"input\": \"competitors capex/EUR in the Delaware Basin\"}\n",
|
||||
"=== Function Output ===\n",
|
||||
"The competitors' Capex/EUR in the Delaware Basin can be found in the image on the slide titled \"Delaware: Vast Inventory with Proven Track Record of Performance.\" The relevant information is presented in a bar chart under the section \"Delaware Basin Well Capex/EUR ($/BOE)\".\n",
|
||||
"\n",
|
||||
"Here are the details:\n",
|
||||
"\n",
|
||||
"- ConocoPhillips: $10/BOE\n",
|
||||
"- Competitor 1: $15/BOE\n",
|
||||
"- Competitor 2: $20/BOE\n",
|
||||
"- Competitor 3: $25/BOE\n",
|
||||
"- Competitor 4: $30/BOE\n",
|
||||
"- Competitor 5: $35/BOE\n",
|
||||
"- Competitor 6: $40/BOE\n",
|
||||
"- Competitor 7: $45/BOE\n",
|
||||
"\n",
|
||||
"This information was obtained directly from the image, which provides a clear visual representation of the Capex/EUR values for ConocoPhillips and its competitors in the Delaware Basin. The parsed markdown text also confirms these values, ensuring consistency between the image and the text.\n",
|
||||
"=== LLM Response ===\n",
|
||||
"The capital expenditure per estimated ultimate recovery (capex/EUR) for ConocoPhillips in the Delaware Basin is $10 per barrel of oil equivalent (BOE). When compared to its competitors, ConocoPhillips has a significantly lower capex/EUR. Here are the capex/EUR values for ConocoPhillips and its competitors:\n",
|
||||
"\n",
|
||||
"- **ConocoPhillips**: $10/BOE\n",
|
||||
"- **Competitor 1**: $15/BOE\n",
|
||||
"- **Competitor 2**: $20/BOE\n",
|
||||
"- **Competitor 3**: $25/BOE\n",
|
||||
"- **Competitor 4**: $30/BOE\n",
|
||||
"- **Competitor 5**: $35/BOE\n",
|
||||
"- **Competitor 6**: $40/BOE\n",
|
||||
"- **Competitor 7**: $45/BOE\n",
|
||||
"\n",
|
||||
"This data indicates that ConocoPhillips has a more cost-efficient operation in the Delaware Basin compared to its competitors.\n",
|
||||
"The capital expenditure per estimated ultimate recovery (capex/EUR) for ConocoPhillips in the Delaware Basin is $10 per barrel of oil equivalent (BOE). When compared to its competitors, ConocoPhillips has a significantly lower capex/EUR. Here are the capex/EUR values for ConocoPhillips and its competitors:\n",
|
||||
"\n",
|
||||
"- **ConocoPhillips**: $10/BOE\n",
|
||||
"- **Competitor 1**: $15/BOE\n",
|
||||
"- **Competitor 2**: $20/BOE\n",
|
||||
"- **Competitor 3**: $25/BOE\n",
|
||||
"- **Competitor 4**: $30/BOE\n",
|
||||
"- **Competitor 5**: $35/BOE\n",
|
||||
"- **Competitor 6**: $40/BOE\n",
|
||||
"- **Competitor 7**: $45/BOE\n",
|
||||
"\n",
|
||||
"This data indicates that ConocoPhillips has a more cost-efficient operation in the Delaware Basin compared to its competitors.\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"# response = agent.query(\"Tell me about the different regions and subregions where Conoco Phillips has a production base.\")\n",
|
||||
"response = agent.query(\n",
|
||||
" \"How does the Conoco Phillips capex/EUR in the delaware basin compare against other competitors?\"\n",
|
||||
")\n",
|
||||
"print(str(response))"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "d584c560-8f49-4c10-a4db-2e0d3b7085d2",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"page_num: 38\n",
|
||||
"image_path: data_images/d9137e19-3974-4b5d-998f-dac0cf29dd9d-page-37.jpg\n",
|
||||
"parsed_text_markdown: # Delaware: Vast Inventory with Proven Track Record of Performance\n",
|
||||
"\n",
|
||||
"## Prolific Acreage Spanning Over ~659,000 Net Acres¹\n",
|
||||
"\n",
|
||||
"\n",
|
||||
"\n",
|
||||
"### Total 10-Year Operated Permian Inventory\n",
|
||||
"\n",
|
||||
"- Delaware Basin: 65%\n",
|
||||
"- Midland Basin: 35%\n",
|
||||
"\n",
|
||||
"### High Single-Digit Production Growth\n",
|
||||
"\n",
|
||||
"## 12-Month Cumulative Production³ (BOE/FT)\n",
|
||||
"\n",
|
||||
"| Months | 2019 | 2020 | 2021 | 2022 |\n",
|
||||
"|--------|------|------|------|------|\n",
|
||||
"| 1 | 0 | 0 | 0 | 0 |\n",
|
||||
"| 2 | 5 | 6 | 7 | 8 |\n",
|
||||
"| 3 | 10 | 12 | 14 | 16 |\n",
|
||||
"| 4 | 15 | 18 | 21 | 24 |\n",
|
||||
"| 5 | 20 | 24 | 28 | 32 |\n",
|
||||
"| 6 | 25 | 30 | 35 | 40 |\n",
|
||||
"| 7 | 30 | 36 | 42 | 48 |\n",
|
||||
"| 8 | 35 | 42 | 49 | 56 |\n",
|
||||
"| 9 | 40 | 48 | 56 | 64 |\n",
|
||||
"| 10 | 45 | 54 | 63 | 72 |\n",
|
||||
"| 11 | 50 | 60 | 70 | 80 |\n",
|
||||
"| 12 | 55 | 66 | 77 | 88 |\n",
|
||||
"\n",
|
||||
"~30% Improved Performance from 2019 to 2022\n",
|
||||
"\n",
|
||||
"## Delaware Basin Well Capex/EUR⁴ ($/BOE)\n",
|
||||
"\n",
|
||||
"| Company | Capex/EUR |\n",
|
||||
"|------------------|-----------|\n",
|
||||
"| ConocoPhillips | 10 |\n",
|
||||
"| Competitor 1 | 15 |\n",
|
||||
"| Competitor 2 | 20 |\n",
|
||||
"| Competitor 3 | 25 |\n",
|
||||
"| Competitor 4 | 30 |\n",
|
||||
"| Competitor 5 | 35 |\n",
|
||||
"| Competitor 6 | 40 |\n",
|
||||
"| Competitor 7 | 45 |\n",
|
||||
"\n",
|
||||
"---\n",
|
||||
"\n",
|
||||
"¹ Unconventional acres. \n",
|
||||
"² Source: Enverus and ConocoPhillips (March 2023). \n",
|
||||
"³ Source: Enverus (March 2023) based on wells online year. \n",
|
||||
"⁴ Source: Enverus (March 2023). Average single well capex/EUR. Top eight public operators based on wells online in years 2021-2022, greater than 50% oil weight. COP based on COP well design. Competitors include: CVX, DVN, EOG, MTDR, OXY, PR and XOM.\n",
|
||||
"parsed_text: \n",
|
||||
"Delaware: Vast Inventory with Proven Track Record of Performance\n",
|
||||
" New Prolific Acreage Spanning Over 12-Month Cumulative Production? (BOE/FT)\n",
|
||||
" Mexico 659,000 Net Acres' 40\n",
|
||||
" Texas 3828\n",
|
||||
" 30 2019\n",
|
||||
" 20 30%\n",
|
||||
" 10 Improved Performancefrom 2019 to 2022\n",
|
||||
" Total\n",
|
||||
" Permian Inventory\n",
|
||||
" 10-Year Operated\n",
|
||||
" 2 10 11 12\n",
|
||||
" Months\n",
|
||||
" Delaware Basin Well Capex/EUR4 (S/BOE)\n",
|
||||
" 65% 25\n",
|
||||
" Delaware Basin 20\n",
|
||||
" Midland Basin 15\n",
|
||||
" Low HighCost of Supplyz 10 ConocoPhillips\n",
|
||||
" High Single-Digit Production Growth\n",
|
||||
" \"Unconventional acres. 2Source: Enverus and ConocoPhillips (March 2023). 3SourceEnverus (March 2023) based on wells online year: \"Source; Enverus (March 2023). Average single well capex/EUR Top eight public operators based on\n",
|
||||
"wells online in years 2021-2022, greater than 50% oil weight; COP based on COP well design: Competitors include; CVX DVN, EOG; MTDR, OXY, PR and XOM: ConocoPhillips\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"print(response.source_nodes[0].get_content(metadata_mode=\"all\"))"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "d21d694b-6618-4d04-a6f6-8b0c2625f539",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"Added user message to memory: How does the Conoco Phillips capex/EUR in the delaware basin compare against other competitors?\n",
|
||||
"=== Calling Function ===\n",
|
||||
"Calling function: vector_tool with args: {\"input\": \"Conoco Phillips capex/EUR in the Delaware Basin\"}\n",
|
||||
"=== Function Output ===\n",
|
||||
"ConocoPhillips' capex/EUR in the Delaware Basin is approximately $20/BOE.\n",
|
||||
"=== Calling Function ===\n",
|
||||
"Calling function: vector_tool with args: {\"input\": \"competitors capex/EUR in the Delaware Basin\"}\n",
|
||||
"=== Function Output ===\n",
|
||||
"The average single well capex/EUR for competitors in the Delaware Basin is between $10 and $25 per BOE.\n",
|
||||
"=== LLM Response ===\n",
|
||||
"ConocoPhillips' capex/EUR in the Delaware Basin is approximately $20 per BOE. In comparison, the average capex/EUR for competitors in the Delaware Basin ranges between $10 and $25 per BOE. This places ConocoPhillips' capex/EUR towards the higher end of the competitive range.\n",
|
||||
"ConocoPhillips' capex/EUR in the Delaware Basin is approximately $20 per BOE. In comparison, the average capex/EUR for competitors in the Delaware Basin ranges between $10 and $25 per BOE. This places ConocoPhillips' capex/EUR towards the higher end of the competitive range.\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"# base_response = base_agent.query(\"Tell me about the different regions and subregions where Conoco Phillips has a production base.\")\n",
|
||||
"base_response = base_agent.query(\n",
|
||||
" \"How does the Conoco Phillips capex/EUR in the delaware basin compare against other competitors?\"\n",
|
||||
")\n",
|
||||
"print(str(base_response))"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "d3afccae-ad8d-4c5d-9d93-810dba413a5d",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"Deep, Durable and Diverse Portfolio with Significant Growth Runway\n",
|
||||
" 1,2002022 Lower 48 Unconventional Production' (MBOED S50 ~S32/BBL\n",
|
||||
" 000 ConocoPhillips Cost of SupplyAverage\n",
|
||||
" 00 S40\n",
|
||||
" 500 3\n",
|
||||
" 400 1 S30\n",
|
||||
" 200\n",
|
||||
" 5\n",
|
||||
" 15,000ConocoPhillipsNet Remaining Well Inventory? 1 S20\n",
|
||||
" 12,000 S10\n",
|
||||
" 000\n",
|
||||
" 0o0 SO\n",
|
||||
" 3,000 10\n",
|
||||
" Resource (BBOE)\n",
|
||||
" Delaware Basin Midland Basin Eagle Ford Bakken Other\n",
|
||||
" Largest Lower 48 Unconventional Producer; Growing into the Next Decade\n",
|
||||
" onshore operated inventory that achieves 15% IRR at $SO/BBL WTI, Competitors include CVX, DVN, EOG, FANG, MRO, OXY, PXD,and XOM:\n",
|
||||
" Source: Wood Mackenzie Lower 48 Unconventional Plays 2022 ProductionCompetitors include CVX, DVN; EOG, FANG, MRO, OXY, PXD and XOM; greaterthan50% liquids weight: ?Source: Wood Mackenzie (March 2023), Lower 48\n",
|
||||
" ConocoPhillips\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"print(base_response.source_nodes[0].get_content(metadata_mode=\"llm\"))"
|
||||
]
|
||||
}
|
||||
],
|
||||
"metadata": {
|
||||
"kernelspec": {
|
||||
"display_name": "llama_index_v3",
|
||||
"language": "python",
|
||||
"name": "llama_index_v3"
|
||||
},
|
||||
"language_info": {
|
||||
"codemirror_mode": {
|
||||
"name": "ipython",
|
||||
"version": 3
|
||||
},
|
||||
"file_extension": ".py",
|
||||
"mimetype": "text/x-python",
|
||||
"name": "python",
|
||||
"nbconvert_exporter": "python",
|
||||
"pygments_lexer": "ipython3"
|
||||
}
|
||||
},
|
||||
"nbformat": 4,
|
||||
"nbformat_minor": 5
|
||||
}
|
||||
Binary file not shown.
|
After Width: | Height: | Size: 271 KiB |
Binary file not shown.
@@ -0,0 +1,359 @@
|
||||
{
|
||||
"cells": [
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "f20600ce-d57a-446e-b033-3aadeec39c1b",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"# LlamaParse with GPT-4o\n",
|
||||
"\n",
|
||||
"\n",
|
||||
"<a href=\"https://colab.research.google.com/github/run-llama/llama_parse/blob/main/examples/test_tesla_impact_report/test_gpt4o.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>\n",
|
||||
"\n",
|
||||
"GPT-4o is a [fully multimodal model by OpenAI](https://openai.com/index/hello-gpt-4o/) released in May 2024. It matches GPT-4 Turbo performance in text and code, and has significantly improved vision and audio capabilities.\n",
|
||||
"\n",
|
||||
"The expanded vision/audio capabilities mean that it can be used for document parsing, by treating each page as an image and performing document extraction. We support using GPT-4o natively in LlamaParse for document parsing. The notebook below walks you through an example of using GPT-4o over the Tesla impact report.\n",
|
||||
"\n",
|
||||
"**NOTE**: The pricing for LlamaParse + gpt4o is an order more expensive than using LlamaParse by default. Currently, every page parsed with gpt4o counts for 10 pages in the LlamaParse usage tracker.\n"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "86b173ac-9fce-4813-bdf1-6dd7d93a491d",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"import nest_asyncio\n",
|
||||
"\n",
|
||||
"nest_asyncio.apply()"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "ecc5eba5-96ce-4db7-bba1-f9ece33e681c",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"import os"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "b805592b-d1a5-4cd2-b916-348f66ca7941",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"os.environ[\"LLAMA_CLOUD_API_KEY\"] = \"<LLAMA_CLOUD_API_KEY>"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "6e73e3c4-9e09-4cba-805f-326c82be812d",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"### Use LlamaParse with `gpt4o_mode=True`\n",
|
||||
"\n",
|
||||
"By turning on gpt4o, we use GPT-4o multimodal capabilities to do document parsing per page instead of the LlamaParse default pipeline.\n",
|
||||
"\n",
|
||||
"We load a snippet of the [2019 Tesla impact report](https://www.tesla.com/ns_videos/2019-tesla-impact-report.pdf). **NOTE**: The report is 57 pages, but will count for 570 pages in LlamaParse due to GPT-4o usage (which is approximately $1.71 USD).\n",
|
||||
"\n",
|
||||
"You can optionally choose to provide a `gpt4o_api_key`. If you do this, then we will use your API key to make GPT-4o calls, and your LlamaParse usage will be counted as if `gpt4o_mode` was not turned on (each page will be counted as a page instead of 10 pages). "
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "aaa2ec5d-f27c-4262-80bf-e57daacff182",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"--2024-05-21 00:10:32-- https://www.dropbox.com/scl/fi/vu6w1dsfo5eddydz13ssm/2019-tesla-impact-report-15.pdf?rlkey=ik8lfqbg2p1ervss4qqt3xose&st=70j04z8j&dl=1\n",
|
||||
"Resolving www.dropbox.com (www.dropbox.com)... 2620:100:6057:18::a27d:d12, 162.125.13.18\n",
|
||||
"Connecting to www.dropbox.com (www.dropbox.com)|2620:100:6057:18::a27d:d12|:443... connected.\n",
|
||||
"HTTP request sent, awaiting response... 302 Found\n",
|
||||
"Location: https://uc872df1ff4ea2fecd3d024fa97a.dl.dropboxusercontent.com/cd/0/inline/CTTnZs8U4V1GtUCNxoB7INwmLq2yU97Q6QbWS6uVnb_XdHe368GrqF0zLDEKTnpc-x7utwNUUpMvWjLyrujrqNVrbGKTKa6hwHu5BxYPA2zXYrzdAEZyeve274xpHZKFywQ/file?dl=1# [following]\n",
|
||||
"--2024-05-21 00:10:33-- https://uc872df1ff4ea2fecd3d024fa97a.dl.dropboxusercontent.com/cd/0/inline/CTTnZs8U4V1GtUCNxoB7INwmLq2yU97Q6QbWS6uVnb_XdHe368GrqF0zLDEKTnpc-x7utwNUUpMvWjLyrujrqNVrbGKTKa6hwHu5BxYPA2zXYrzdAEZyeve274xpHZKFywQ/file?dl=1\n",
|
||||
"Resolving uc872df1ff4ea2fecd3d024fa97a.dl.dropboxusercontent.com (uc872df1ff4ea2fecd3d024fa97a.dl.dropboxusercontent.com)... 2620:100:6057:15::a27d:d0f, 162.125.13.15\n",
|
||||
"Connecting to uc872df1ff4ea2fecd3d024fa97a.dl.dropboxusercontent.com (uc872df1ff4ea2fecd3d024fa97a.dl.dropboxusercontent.com)|2620:100:6057:15::a27d:d0f|:443... connected.\n",
|
||||
"HTTP request sent, awaiting response... 302 Found\n",
|
||||
"Location: /cd/0/inline2/CTSaARDHbxvyEEgefshmsHLbuXkgV1Rmr-ItVhk5lPuZXkLlNnZMZWCF9YF5j4t2lLs4VurFW2VI1Q4A6ZFi8D2RXJmUG3wdgJhR6qSaBpwRZxjB_vk8qkJb8h1jRDaL7ATK6XYTHncab_aoPWzB62vrZ9yXUM0Mr-EdCX1k-hMbzXLV2dorA71IuFPY8ICkTemRWaG6VhBd3bV0C5AkMsAqy90w6Kez1ySFO06UkrxLSmkCaKdFgVoLcUVO2PLv4rGv6AuZOF_kqwsHdh82J9DQU4PMMyg-f5ChSGGSCKgmUfTBE2qP1eISP-GXSB91yWwMf-7rxGtM8MpDp-AL5jxYZxhZcmZn1cU8Or_8OOZrxg/file?dl=1 [following]\n",
|
||||
"--2024-05-21 00:10:33-- https://uc872df1ff4ea2fecd3d024fa97a.dl.dropboxusercontent.com/cd/0/inline2/CTSaARDHbxvyEEgefshmsHLbuXkgV1Rmr-ItVhk5lPuZXkLlNnZMZWCF9YF5j4t2lLs4VurFW2VI1Q4A6ZFi8D2RXJmUG3wdgJhR6qSaBpwRZxjB_vk8qkJb8h1jRDaL7ATK6XYTHncab_aoPWzB62vrZ9yXUM0Mr-EdCX1k-hMbzXLV2dorA71IuFPY8ICkTemRWaG6VhBd3bV0C5AkMsAqy90w6Kez1ySFO06UkrxLSmkCaKdFgVoLcUVO2PLv4rGv6AuZOF_kqwsHdh82J9DQU4PMMyg-f5ChSGGSCKgmUfTBE2qP1eISP-GXSB91yWwMf-7rxGtM8MpDp-AL5jxYZxhZcmZn1cU8Or_8OOZrxg/file?dl=1\n",
|
||||
"Reusing existing connection to [uc872df1ff4ea2fecd3d024fa97a.dl.dropboxusercontent.com]:443.\n",
|
||||
"HTTP request sent, awaiting response... 200 OK\n",
|
||||
"Length: 26199694 (25M) [application/binary]\n",
|
||||
"Saving to: ‘2019-tesla-impact-report-15.pdf’\n",
|
||||
"\n",
|
||||
"2019-tesla-impact-r 100%[===================>] 24.99M 30.5MB/s in 0.8s \n",
|
||||
"\n",
|
||||
"2024-05-21 00:10:35 (30.5 MB/s) - ‘2019-tesla-impact-report-15.pdf’ saved [26199694/26199694]\n",
|
||||
"\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"!wget \"https://www.dropbox.com/scl/fi/vu6w1dsfo5eddydz13ssm/2019-tesla-impact-report-15.pdf?rlkey=ik8lfqbg2p1ervss4qqt3xose&st=70j04z8j&dl=1\" -O \"2019-tesla-impact-report-15.pdf\""
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "f46991c1-031b-461f-b9a6-9237a821f4c8",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"from llama_parse import LlamaParse\n",
|
||||
"\n",
|
||||
"parser_gpt4o = LlamaParse(\n",
|
||||
" result_type=\"markdown\",\n",
|
||||
" # api_key=api_key,\n",
|
||||
" gpt4o_mode=True,\n",
|
||||
" split_by_page=True,\n",
|
||||
" # gpt4o_api_key=\"<gpt4o_api_key>\"\n",
|
||||
")"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "1136ba82-074b-489d-9b0a-d609ccbf02b6",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"Started parsing the file under job_id bf7d4619-3e26-479d-80e9-25702186ef32\n",
|
||||
"."
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"documents_gpt4o = parser_gpt4o.load_data(\"./2019-tesla-impact-report-15.pdf\")"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "9e65c54f-3e4c-4c78-b1e8-a55ebeba1f24",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"# Mission & Tesla Ecosystem\n",
|
||||
"\n",
|
||||
"Climate change is reaching alarming levels in large part due to emissions from burning fossil fuels for transportation and electricity generation. In 2016, carbon dioxide (CO2) concentration levels in the atmosphere exceeded the 400 parts per million threshold on a sustained basis - a level that climate scientists believe will have a catastrophic impact on the environment. Worse, annual global CO2 emissions continue to increase and have approximately doubled over the past 50 years to over 43 gigatons in 2019. The world’s current path is unwise and unsustainable.\n",
|
||||
"\n",
|
||||
"The world cannot reduce CO2 emissions without addressing both energy generation and consumption. And the world cannot address its energy habits without first directly reducing emissions in the transportation and energy sectors. We are focused on creating a complete energy and transportation ecosystem from solar generation and energy storage to all-electric vehicles that produce zero tailpipe emissions.\n",
|
||||
"\n",
|
||||
"Since the onset of shelter-in-place orders and travel restrictions due to COVID-19, we have seen dramatic increases in air quality across the planet, as well as projections for CO2 emissions to drop in excess of 4% in 2020 compared to pre-COVID-19 levels, according to researchers. Because these improvements in air quality and reductions in CO2 are a result of a global economic disruption and not due to systemic changes in how we produce and consume energy, they are not expected to be sustained absent intervention. However, these changes have shown us the positive impacts of reduced pollution in a very short period of time. At Tesla, we believe that we all have an unprecedented opportunity to learn from this disruption and accelerate the deployment of clean energy solutions as part of a recovery for all economies throughout the world, and we will actively continue to advocate for the realization of these long-term changes.\n",
|
||||
"\n",
|
||||
"| Global Greenhouse Gas (GHG) Emissions by Economic Sector |\n",
|
||||
"|----------------------------------------------------------|\n",
|
||||
"|  |\n",
|
||||
"\n",
|
||||
"| Sector | Percentage |\n",
|
||||
"|---------------------------------------------|------------|\n",
|
||||
"| Electricity & Heat Production* | 31% |\n",
|
||||
"| Agriculture, Forestry & Other Land Use | 20% |\n",
|
||||
"| Industry | 18% |\n",
|
||||
"| Transportation* | 16% |\n",
|
||||
"| Other Energy | 9% |\n",
|
||||
"| Buildings | 6% |\n",
|
||||
"\n",
|
||||
"*Tesla-related sectors. Source: World Resources Institute\n",
|
||||
"\n",
|
||||
"According to the Global Carbon project, when fully tallied, total carbon emissions from 2019 are expected to hit another record high of over 43 gigatons for the year. Energy use through electricity and heat production (31%) and transportation (16%) are significant drivers of these GHG emissions.\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"print(documents_gpt4o[3].get_content())"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "d62cbb62-37ea-4370-9411-d979aa3a627e",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## Build RAG pipeline over the Parsed Report\n",
|
||||
"\n",
|
||||
"We now try building a RAG pipeline over this parsed report. It's not a lot of text, but we split it into chunks and load it into a simple in-memory vector store.\n",
|
||||
"\n",
|
||||
"We ask a question over the parsed markdown table and get back the right answer! We also ask a question over the text."
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "d8b7c3ad-2147-448c-bcbe-3e6fcd8d5361",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"from llama_index.core import VectorStoreIndex\n",
|
||||
"\n",
|
||||
"vector_index = VectorStoreIndex(documents_gpt4o)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "8013351a-180d-4947-9f81-513042175c19",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"query_engine = vector_index.as_query_engine(similarity_top_k=6)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "795dc5c4-e122-4ff3-94d2-747fa51d5add",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"response = query_engine.query(\n",
|
||||
" \"What are the greenhouse emissions for agriculture and transportation?\"\n",
|
||||
")"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "39d2e6bd-3316-49b5-9a5d-5b4b95343e5a",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"Agriculture accounts for 20% of global greenhouse gas emissions, while transportation contributes 16% of these emissions.\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"print(str(response))"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"id": "9beb5cd4-4041-48c7-b22b-de5540f92a6d",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"Let's also try asking a question over another piece of the text."
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "543c8b63-5cd1-47a1-a8a1-81abbfd3e52b",
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"response = query_engine.query(\n",
|
||||
" \"How does the EPA range of Teslas compare with other vehicles? Give details\"\n",
|
||||
")"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "e739eabf-732b-4f59-9628-972c4bf6c857",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"The EPA range of Tesla vehicles varies across different models. The Model 3 Standard Range Plus (SR+) achieves an EPA range of 4.8 miles/kWh, making it the most efficient electric vehicle in production. The Model Y all-wheel drive (AWD) achieves 4.1 miles/kWh, which positions it as the most efficient electric SUV produced to date. The energy efficiency of Tesla vehicles is highlighted by these EPA range figures, showcasing their advancements in powertrain efficiency compared to other electric vehicles on the market.\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"print(str(response))"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"id": "04b05c53-1a81-41a7-97f2-98a960211957",
|
||||
"metadata": {},
|
||||
"outputs": [
|
||||
{
|
||||
"name": "stdout",
|
||||
"output_type": "stream",
|
||||
"text": [
|
||||
"# Reducing Carbon Footprint Even Further\n",
|
||||
"## Improving Powertrain Efficiency\n",
|
||||
"\n",
|
||||
"Tesla vehicles are known to have the highest energy efficiency of any EV built to date. In the early days of Model S production, we were able to achieve energy efficiency of 3.1 EPA miles / kWh. Today, our most efficient Model 3 Standard Range Plus (SR+) achieves an EPA range of 4.8 miles / kWh, more than any EV in production. Model Y all-wheel drive (AWD) achieves 4.1 EPA miles / kWh, which makes it the most efficient electric SUV produced to date.\n",
|
||||
"\n",
|
||||
"The energy efficiency of Tesla vehicles will continue to improve further over time as we continue to improve our technology and powertrain efficiency. It is also reasonable to assume that our high-mileage products, such as our future Tesla Robotaxis, will be designed for maximum energy efficiency as handling, acceleration, and top speed become less relevant. That way, we will minimize cost for our customers as well as reduce the carbon footprint per mile driven.\n",
|
||||
"\n",
|
||||
"### Average Lifecycle Emissions in U.S. (gCO2e/mi)\n",
|
||||
"\n",
|
||||
"| Vehicle Type | Manufacturing Phase | Use Phase | Total Emissions |\n",
|
||||
"|---------------------------------------|---------------------|-----------|-----------------|\n",
|
||||
"| Avg. Mid-Size Premium ICE | | | |\n",
|
||||
"| Model 3 Personal Use (grid charged) | | | |\n",
|
||||
"| Model 3 Ridesharing Use (grid charged)| | | |\n",
|
||||
"| Model 3 Personal Use (solar charged) | | | |\n",
|
||||
"| Model 3 Ridesharing Use (solar charged)| | | |\n",
|
||||
"\n",
|
||||
"*Note: The chart shows that the emissions depend on powertrain efficiency.*\n",
|
||||
"\n",
|
||||
"### Energy Efficiency EPA range in miles/kWh\n",
|
||||
"\n",
|
||||
"| Vehicle Model | EPA Range (miles/kWh) |\n",
|
||||
"|---------------------|-----------------------|\n",
|
||||
"| Model 3 SR+ | 4.8 |\n",
|
||||
"| Model 3 AWD | |\n",
|
||||
"| Model Y AWD | |\n",
|
||||
"| Hyundai Kona | |\n",
|
||||
"| Chevy Bolt | |\n",
|
||||
"| Model S LR+ | |\n",
|
||||
"| Nissan Leaf | |\n",
|
||||
"| Model X LR+ | |\n",
|
||||
"| Jaguar iPace | |\n",
|
||||
"| Mercedes EQC* | |\n",
|
||||
"| Ford Mach E AWD | |\n",
|
||||
"| Audi e-tron | |\n",
|
||||
"| Porsche Taycan | |\n",
|
||||
"\n",
|
||||
"*Tesla estimate. Source: OEM websites*\n"
|
||||
]
|
||||
}
|
||||
],
|
||||
"source": [
|
||||
"print(response.source_nodes[0].get_content())"
|
||||
]
|
||||
}
|
||||
],
|
||||
"metadata": {
|
||||
"kernelspec": {
|
||||
"display_name": "llama_parse",
|
||||
"language": "python",
|
||||
"name": "llama_parse"
|
||||
},
|
||||
"language_info": {
|
||||
"codemirror_mode": {
|
||||
"name": "ipython",
|
||||
"version": 3
|
||||
},
|
||||
"file_extension": ".py",
|
||||
"mimetype": "text/x-python",
|
||||
"name": "python",
|
||||
"nbconvert_exporter": "python",
|
||||
"pygments_lexer": "ipython3"
|
||||
}
|
||||
},
|
||||
"nbformat": 4,
|
||||
"nbformat_minor": 5
|
||||
}
|
||||
+144
-34
@@ -5,6 +5,7 @@ import mimetypes
|
||||
import time
|
||||
from pathlib import Path
|
||||
from typing import List, Optional, Union
|
||||
from io import BufferedIOBase
|
||||
|
||||
from llama_index.core.async_utils import run_jobs
|
||||
from llama_index.core.bridge.pydantic import Field, validator
|
||||
@@ -18,6 +19,26 @@ from llama_parse.utils import (
|
||||
Language,
|
||||
SUPPORTED_FILE_TYPES,
|
||||
)
|
||||
from copy import deepcopy
|
||||
|
||||
# can put in a path to the file or the file bytes itself
|
||||
# if passing as bytes or a buffer, must provide the file_name in extra_info
|
||||
FileInput = Union[str, bytes, BufferedIOBase]
|
||||
|
||||
|
||||
def _get_sub_docs(docs: List[Document]) -> List[Document]:
|
||||
"""Split docs into pages, by separator."""
|
||||
sub_docs = []
|
||||
for doc in docs:
|
||||
doc_chunks = doc.text.split("\n---\n")
|
||||
for doc_chunk in doc_chunks:
|
||||
sub_doc = Document(
|
||||
text=doc_chunk,
|
||||
metadata=deepcopy(doc.metadata),
|
||||
)
|
||||
sub_docs.append(sub_doc)
|
||||
|
||||
return sub_docs
|
||||
|
||||
|
||||
class LlamaParse(BasePydanticReader):
|
||||
@@ -57,6 +78,30 @@ class LlamaParse(BasePydanticReader):
|
||||
parsing_instruction: Optional[str] = Field(
|
||||
default="", description="The parsing instruction for the parser."
|
||||
)
|
||||
skip_diagonal_text: Optional[bool] = Field(
|
||||
default=False,
|
||||
description="If set to true, the parser will ignore diagonal text (when the text rotation in degrees modulo 90 is not 0).",
|
||||
)
|
||||
invalidate_cache: Optional[bool] = Field(
|
||||
default=False,
|
||||
description="If set to true, the cache will be ignored and the document re-processes. All document are kept in cache for 48hours after the job was completed to avoid processing the same document twice.",
|
||||
)
|
||||
do_not_cache: Optional[bool] = Field(
|
||||
default=False,
|
||||
description="If set to true, the document will not be cached. This mean that you will be re-charged it you reprocess them as they will not be cached.",
|
||||
)
|
||||
fast_mode: Optional[bool] = Field(
|
||||
default=False,
|
||||
description="Note: Non compatible with gpt-4o. If set to true, the parser will use a faster mode to extract text from documents. This mode will skip OCR of images, and table/heading reconstruction.",
|
||||
)
|
||||
do_not_unroll_columns: Optional[bool] = Field(
|
||||
default=False,
|
||||
description="If set to true, the parser will keep column in the text according to document layout. Reduce reconstruction accuracy, and LLM's/embedings performances in most case.",
|
||||
)
|
||||
page_separator: Optional[str] = Field(
|
||||
default=None,
|
||||
description="The page separator to use to split the text. Default is None, which means the parser will use the default separator '\\n---\\n'.",
|
||||
)
|
||||
gpt4o_mode: bool = Field(
|
||||
default=False,
|
||||
description="Whether to use gpt-4o extract text from documents.",
|
||||
@@ -65,10 +110,34 @@ class LlamaParse(BasePydanticReader):
|
||||
default=None,
|
||||
description="The API key for the GPT-4o API. Lowers the cost of parsing.",
|
||||
)
|
||||
bounding_box: Optional[str] = Field(
|
||||
default=None,
|
||||
description="The bounding box to use to extract text from documents describe as a string containing the bounding box margins",
|
||||
)
|
||||
target_pages: Optional[str] = Field(
|
||||
default=None,
|
||||
description="The target pages to extract text from documents. Describe as a comma separated list of page numbers. The first page of the document is page 0",
|
||||
)
|
||||
ignore_errors: bool = Field(
|
||||
default=True,
|
||||
description="Whether or not to ignore and skip errors raised during parsing.",
|
||||
)
|
||||
split_by_page: bool = Field(
|
||||
default=True,
|
||||
description="Whether to split by page (NOTE: using a predefined separator `\n---\n`)",
|
||||
)
|
||||
vendor_multimodal_api_key: Optional[str] = Field(
|
||||
default=None,
|
||||
description="The API key for the multimodal API.",
|
||||
)
|
||||
use_vendor_multimodal_model: bool = Field(
|
||||
default=False,
|
||||
description="Whether to use the vendor multimodal API.",
|
||||
)
|
||||
vendor_multimodal_model_name: Optional[str] = Field(
|
||||
default=None,
|
||||
description="The model name for the vendor multimodal API.",
|
||||
)
|
||||
|
||||
@validator("api_key", pre=True, always=True)
|
||||
def validate_api_key(cls, v: str) -> str:
|
||||
@@ -91,28 +160,39 @@ class LlamaParse(BasePydanticReader):
|
||||
|
||||
# upload a document and get back a job_id
|
||||
async def _create_job(
|
||||
self, file_path: str, extra_info: Optional[dict] = None
|
||||
self, file_input: FileInput, extra_info: Optional[dict] = None
|
||||
) -> str:
|
||||
file_path = str(file_path)
|
||||
file_ext = os.path.splitext(file_path)[1]
|
||||
if file_ext not in SUPPORTED_FILE_TYPES:
|
||||
raise Exception(
|
||||
f"Currently, only the following file types are supported: {SUPPORTED_FILE_TYPES}\n"
|
||||
f"Current file type: {file_ext}"
|
||||
headers = {"Authorization": f"Bearer {self.api_key}"}
|
||||
url = f"{self.base_url}/api/parsing/upload"
|
||||
files = None
|
||||
file_handle = None
|
||||
|
||||
if isinstance(file_input, (bytes, BufferedIOBase)):
|
||||
if not extra_info or "file_name" not in extra_info:
|
||||
raise ValueError(
|
||||
"file_name must be provided in extra_info when passing bytes"
|
||||
)
|
||||
file_name = extra_info["file_name"]
|
||||
mime_type = mimetypes.guess_type(file_name)[0]
|
||||
files = {"file": (file_name, file_input, mime_type)}
|
||||
elif isinstance(file_input, (str, Path)):
|
||||
file_path = str(file_input)
|
||||
file_ext = os.path.splitext(file_path)[1].lower()
|
||||
if file_ext not in SUPPORTED_FILE_TYPES:
|
||||
raise Exception(
|
||||
f"Currently, only the following file types are supported: {SUPPORTED_FILE_TYPES}\n"
|
||||
f"Current file type: {file_ext}"
|
||||
)
|
||||
mime_type = mimetypes.guess_type(file_path)[0]
|
||||
# Open the file here for the duration of the async context
|
||||
file_handle = open(file_path, "rb")
|
||||
files = {"file": (os.path.basename(file_path), file_handle, mime_type)}
|
||||
else:
|
||||
raise ValueError(
|
||||
"file_input must be either a file path string, file bytes, or buffer object"
|
||||
)
|
||||
|
||||
extra_info = extra_info or {}
|
||||
extra_info["file_path"] = file_path
|
||||
|
||||
headers = {"Authorization": f"Bearer {self.api_key}"}
|
||||
|
||||
# load data, set the mime type
|
||||
with open(file_path, "rb") as f:
|
||||
mime_type = mimetypes.guess_type(file_path)[0]
|
||||
files = {"file": (f.name, f, mime_type)}
|
||||
|
||||
# send the request, start job
|
||||
url = f"{self.base_url}/api/parsing/upload"
|
||||
try:
|
||||
async with httpx.AsyncClient(timeout=self.max_timeout) as client:
|
||||
response = await client.post(
|
||||
url,
|
||||
@@ -121,16 +201,28 @@ class LlamaParse(BasePydanticReader):
|
||||
data={
|
||||
"language": self.language.value,
|
||||
"parsing_instruction": self.parsing_instruction,
|
||||
"invalidate_cache": self.invalidate_cache,
|
||||
"skip_diagonal_text": self.skip_diagonal_text,
|
||||
"do_not_cache": self.do_not_cache,
|
||||
"fast_mode": self.fast_mode,
|
||||
"do_not_unroll_columns": self.do_not_unroll_columns,
|
||||
"page_separator": self.page_separator,
|
||||
"gpt4o_mode": self.gpt4o_mode,
|
||||
"gpt4o_api_key": self.gpt4o_api_key,
|
||||
"bounding_box": self.bounding_box,
|
||||
"target_pages": self.target_pages,
|
||||
"vendor_multimodal_api_key": self.vendor_multimodal_api_key,
|
||||
"use_vendor_multimodal_model": self.use_vendor_multimodal_model,
|
||||
"vendor_multimodal_model_name": self.vendor_multimodal_model_name,
|
||||
},
|
||||
)
|
||||
if not response.is_success:
|
||||
raise Exception(f"Failed to parse the file: {response.text}")
|
||||
|
||||
# check the status of the job, return when done
|
||||
job_id = response.json()["id"]
|
||||
return job_id
|
||||
job_id = response.json()["id"]
|
||||
return job_id
|
||||
finally:
|
||||
if file_handle is not None:
|
||||
file_handle.close()
|
||||
|
||||
async def _get_job_result(
|
||||
self, job_id: str, result_type: str, verbose: bool = False
|
||||
@@ -180,7 +272,10 @@ class LlamaParse(BasePydanticReader):
|
||||
)
|
||||
|
||||
async def _aload_data(
|
||||
self, file_path: str, extra_info: Optional[dict] = None, verbose: bool = False
|
||||
self,
|
||||
file_path: FileInput,
|
||||
extra_info: Optional[dict] = None,
|
||||
verbose: bool = False,
|
||||
) -> List[Document]:
|
||||
"""Load data from the input path."""
|
||||
try:
|
||||
@@ -192,25 +287,32 @@ class LlamaParse(BasePydanticReader):
|
||||
job_id, self.result_type.value, verbose=verbose
|
||||
)
|
||||
|
||||
return [
|
||||
docs = [
|
||||
Document(
|
||||
text=result[self.result_type.value],
|
||||
metadata=extra_info or {},
|
||||
)
|
||||
]
|
||||
if self.split_by_page:
|
||||
return _get_sub_docs(docs)
|
||||
else:
|
||||
return docs
|
||||
|
||||
except Exception as e:
|
||||
print(f"Error while parsing the file '{file_path}':", e)
|
||||
file_repr = file_path if isinstance(file_path, str) else "<bytes/buffer>"
|
||||
print(f"Error while parsing the file '{file_repr}':", e)
|
||||
if self.ignore_errors:
|
||||
return []
|
||||
else:
|
||||
raise e
|
||||
|
||||
async def aload_data(
|
||||
self, file_path: Union[List[str], str], extra_info: Optional[dict] = None
|
||||
self,
|
||||
file_path: Union[List[FileInput], FileInput],
|
||||
extra_info: Optional[dict] = None,
|
||||
) -> List[Document]:
|
||||
"""Load data from the input path."""
|
||||
if isinstance(file_path, (str, Path)):
|
||||
if isinstance(file_path, (str, Path, bytes, BufferedIOBase)):
|
||||
return await self._aload_data(
|
||||
file_path, extra_info=extra_info, verbose=self.verbose
|
||||
)
|
||||
@@ -244,7 +346,9 @@ class LlamaParse(BasePydanticReader):
|
||||
)
|
||||
|
||||
def load_data(
|
||||
self, file_path: Union[List[str], str], extra_info: Optional[dict] = None
|
||||
self,
|
||||
file_path: Union[List[FileInput], FileInput],
|
||||
extra_info: Optional[dict] = None,
|
||||
) -> List[Document]:
|
||||
"""Load data from the input path."""
|
||||
try:
|
||||
@@ -256,7 +360,7 @@ class LlamaParse(BasePydanticReader):
|
||||
raise e
|
||||
|
||||
async def _aget_json(
|
||||
self, file_path: str, extra_info: Optional[dict] = None
|
||||
self, file_path: FileInput, extra_info: Optional[dict] = None
|
||||
) -> List[dict]:
|
||||
"""Load data from the input path."""
|
||||
try:
|
||||
@@ -270,14 +374,17 @@ class LlamaParse(BasePydanticReader):
|
||||
return [result]
|
||||
|
||||
except Exception as e:
|
||||
print(f"Error while parsing the file '{file_path}':", e)
|
||||
file_repr = file_path if isinstance(file_path, str) else "<bytes/buffer>"
|
||||
print(f"Error while parsing the file '{file_repr}':", e)
|
||||
if self.ignore_errors:
|
||||
return []
|
||||
else:
|
||||
raise e
|
||||
|
||||
async def aget_json(
|
||||
self, file_path: Union[List[str], str], extra_info: Optional[dict] = None
|
||||
self,
|
||||
file_path: Union[List[FileInput], FileInput],
|
||||
extra_info: Optional[dict] = None,
|
||||
) -> List[dict]:
|
||||
"""Load data from the input path."""
|
||||
if isinstance(file_path, (str, Path)):
|
||||
@@ -305,7 +412,9 @@ class LlamaParse(BasePydanticReader):
|
||||
)
|
||||
|
||||
def get_json_result(
|
||||
self, file_path: Union[List[str], str], extra_info: Optional[dict] = None
|
||||
self,
|
||||
file_path: Union[List[FileInput], FileInput],
|
||||
extra_info: Optional[dict] = None,
|
||||
) -> List[dict]:
|
||||
"""Parse the input path."""
|
||||
try:
|
||||
@@ -341,7 +450,8 @@ class LlamaParse(BasePydanticReader):
|
||||
|
||||
# get a valid image path
|
||||
if not image_path.endswith(".png"):
|
||||
image_path += ".png"
|
||||
if not image_path.endswith(".jpg"):
|
||||
image_path += ".png"
|
||||
|
||||
image["path"] = image_path
|
||||
image["job_id"] = job_id
|
||||
|
||||
+76
-35
@@ -101,52 +101,93 @@ class Language(str, Enum):
|
||||
|
||||
SUPPORTED_FILE_TYPES = [
|
||||
".pdf",
|
||||
# Microsoft word - all versions
|
||||
# document and presentations
|
||||
".602",
|
||||
".abw",
|
||||
".cgm",
|
||||
".cwk",
|
||||
".doc",
|
||||
".docx",
|
||||
".docm",
|
||||
".dot",
|
||||
".dotx",
|
||||
".dotm",
|
||||
# Rich text format
|
||||
".hwp",
|
||||
".key",
|
||||
".lwp",
|
||||
".mw",
|
||||
".mcw",
|
||||
".pages",
|
||||
".pbd",
|
||||
".ppt",
|
||||
".pptm",
|
||||
".pptx",
|
||||
".pot",
|
||||
".potm",
|
||||
".potx",
|
||||
".rtf",
|
||||
# Microsoft Works
|
||||
".wps",
|
||||
# Word Perfect
|
||||
".wpd",
|
||||
# Open Office
|
||||
".sda",
|
||||
".sdd",
|
||||
".sdp",
|
||||
".sdw",
|
||||
".sgl",
|
||||
".sti",
|
||||
".sxi",
|
||||
".sxw",
|
||||
".stw",
|
||||
".sxg",
|
||||
# Apple
|
||||
".pages",
|
||||
# Mac Write
|
||||
".mw",
|
||||
".mcw",
|
||||
# Unified Office Format text
|
||||
".uot",
|
||||
".txt",
|
||||
".uof",
|
||||
".uos",
|
||||
".uop",
|
||||
# Microsoft powerpoints
|
||||
".ppt",
|
||||
".pptx",
|
||||
".pot",
|
||||
".pptm",
|
||||
".potx",
|
||||
".potm",
|
||||
# Apple keynote
|
||||
".key",
|
||||
# Open Office Presentations
|
||||
".odp",
|
||||
".odg",
|
||||
".otp",
|
||||
".fopd",
|
||||
".sxi",
|
||||
".sti",
|
||||
# ebook
|
||||
".uot",
|
||||
".vor",
|
||||
".wpd",
|
||||
".wps",
|
||||
".xml",
|
||||
".zabw",
|
||||
".epub",
|
||||
# html
|
||||
".html",
|
||||
# images
|
||||
".jpg",
|
||||
".jpeg",
|
||||
".png",
|
||||
".gif",
|
||||
".bmp",
|
||||
".svg",
|
||||
".tiff",
|
||||
".webp",
|
||||
# web
|
||||
".htm",
|
||||
".html",
|
||||
# spreadsheets
|
||||
".xlsx",
|
||||
".xls",
|
||||
".xlsm",
|
||||
".xlsb",
|
||||
".xlw",
|
||||
".csv",
|
||||
".dif",
|
||||
".sylk",
|
||||
".slk",
|
||||
".prn",
|
||||
".numbers",
|
||||
".et",
|
||||
".ods",
|
||||
".fods",
|
||||
".uos1",
|
||||
".uos2",
|
||||
".dbf",
|
||||
".wk1",
|
||||
".wk2",
|
||||
".wk3",
|
||||
".wk4",
|
||||
".wks",
|
||||
".123",
|
||||
".wq1",
|
||||
".wq2",
|
||||
".wb1",
|
||||
".wb2",
|
||||
".wb3",
|
||||
".qpw",
|
||||
".xlr",
|
||||
".eth",
|
||||
".tsv",
|
||||
]
|
||||
|
||||
Generated
+694
-691
File diff suppressed because it is too large
Load Diff
+1
-1
@@ -4,7 +4,7 @@ build-backend = "poetry.core.masonry.api"
|
||||
|
||||
[tool.poetry]
|
||||
name = "llama-parse"
|
||||
version = "0.4.3"
|
||||
version = "0.4.8"
|
||||
description = "Parse files into RAG-Optimized formats."
|
||||
authors = ["Logan Markewich <logan@llamaindex.ai>"]
|
||||
license = "MIT"
|
||||
|
||||
+43
-7
@@ -18,21 +18,57 @@ def test_simple_page_text() -> None:
|
||||
assert len(result[0].text) > 0
|
||||
|
||||
|
||||
@pytest.mark.skipif(
|
||||
os.environ.get("LLAMA_CLOUD_API_KEY", "") == "",
|
||||
reason="LLAMA_CLOUD_API_KEY not set",
|
||||
)
|
||||
def test_simple_page_markdown() -> None:
|
||||
parser = LlamaParse(result_type="markdown")
|
||||
@pytest.fixture
|
||||
def markdown_parser() -> LlamaParse:
|
||||
if os.environ.get("LLAMA_CLOUD_API_KEY", "") == "":
|
||||
pytest.skip("LLAMA_CLOUD_API_KEY not set")
|
||||
return LlamaParse(result_type="markdown", ignore_errors=False)
|
||||
|
||||
|
||||
def test_simple_page_markdown(markdown_parser: LlamaParse) -> None:
|
||||
filepath = os.path.join(
|
||||
os.path.dirname(__file__), "test_files/attention_is_all_you_need.pdf"
|
||||
)
|
||||
result = markdown_parser.load_data(filepath)
|
||||
assert len(result) == 1
|
||||
assert len(result[0].text) > 0
|
||||
|
||||
|
||||
def test_simple_page_markdown_bytes(markdown_parser: LlamaParse) -> None:
|
||||
markdown_parser = LlamaParse(result_type="markdown", ignore_errors=False)
|
||||
|
||||
filepath = os.path.join(
|
||||
os.path.dirname(__file__), "test_files/attention_is_all_you_need.pdf"
|
||||
)
|
||||
result = parser.load_data(filepath)
|
||||
with open(filepath, "rb") as f:
|
||||
file_bytes = f.read()
|
||||
# client must provide extra_info with file_name
|
||||
with pytest.raises(ValueError):
|
||||
result = markdown_parser.load_data(file_bytes)
|
||||
result = markdown_parser.load_data(
|
||||
file_bytes, extra_info={"file_name": "attention_is_all_you_need.pdf"}
|
||||
)
|
||||
assert len(result) == 1
|
||||
assert len(result[0].text) > 0
|
||||
|
||||
|
||||
def test_simple_page_markdown_buffer(markdown_parser: LlamaParse) -> None:
|
||||
markdown_parser = LlamaParse(result_type="markdown", ignore_errors=False)
|
||||
|
||||
filepath = os.path.join(
|
||||
os.path.dirname(__file__), "test_files/attention_is_all_you_need.pdf"
|
||||
)
|
||||
with open(filepath, "rb") as f:
|
||||
# client must provide extra_info with file_name
|
||||
with pytest.raises(ValueError):
|
||||
result = markdown_parser.load_data(f)
|
||||
result = markdown_parser.load_data(
|
||||
f, extra_info={"file_name": "attention_is_all_you_need.pdf"}
|
||||
)
|
||||
assert len(result) == 1
|
||||
assert len(result[0].text) > 0
|
||||
|
||||
|
||||
@pytest.mark.skipif(
|
||||
os.environ.get("LLAMA_CLOUD_API_KEY", "") == "",
|
||||
reason="LLAMA_CLOUD_API_KEY not set",
|
||||
|
||||
Reference in New Issue
Block a user