run-llama/llama_cloud_services
1
0
Fork 0
mirror of https://github.com/run-llama/llama_cloud_services.git synced 2026-08-26 16:56:36 -04:00
Code Issues 352 Packages Projects Releases 10 Wiki Activity

Issue with LlamaParse content_guideline_instruction #431

New Issue
Closed
opened 2026-02-16 00:17:49 -05:00 by yindo · 5 comments
yindo commented 2026-02-16 00:17:49 -05:00
Owner
Copy Link
Copy Source

Originally created by @mehwishh247 on GitHub (Feb 15, 2025).

Describe the bug
While running LlamaParse library through a Python script, the code just stopped reading the content_guideline_instruction. The same instructions work using the web GUI, but when initiated through Python script, it doesn't work

Files

medical_data.pdf

Job ID
2c92a6ea-42d4-46ac-83ca-afb9d7250aee
390c00c8-26d6-4469-a8de-4169ce229fa4
bf987cc7-226e-4ddf-bccb-bdf144ede533
0b183e90-6f4b-47b9-9349-9afec1bf56df

Client:
Please remove untested options:

  • Python Library
  • Frontend (cloud.llamaindex.ai)

Additional context
I used in default mode (balanced or accurate). I set the content_guideline_instruction instead of parsing_instructions since it was deprecated. Just in case, I tested the code with parsing instructions as well, but it still didn't work. I checked the job history and it just isn't reading the guidelines. After trying for at least 10 time, only one job actually read the instructions as template. I tried again immediately and it didn't work. Apart from instructions, I used default settings for everything.

Image
Image
Image
Image
Image
Image

Originally created by @mehwishh247 on GitHub (Feb 15, 2025). **Describe the bug** While running LlamaParse library through a Python script, the code just stopped reading the **content_guideline_instruction**. The same instructions work using the web GUI, but when initiated through Python script, it doesn't work **Files** [medical_data.pdf](https://github.com/user-attachments/files/18810989/medical_data.pdf) **Job ID** 2c92a6ea-42d4-46ac-83ca-afb9d7250aee 390c00c8-26d6-4469-a8de-4169ce229fa4 bf987cc7-226e-4ddf-bccb-bdf144ede533 0b183e90-6f4b-47b9-9349-9afec1bf56df **Client:** Please remove untested options: - Python Library - Frontend (cloud.llamaindex.ai) **Additional context** I used in default mode (balanced or accurate). I set the content_guideline_instruction instead of parsing_instructions since it was deprecated. Just in case, I tested the code with parsing instructions as well, but it still didn't work. I checked the job history and it just isn't reading the guidelines. After trying for at least 10 time, only one job actually read the instructions as **template**. I tried again immediately and it didn't work. Apart from instructions, I used default settings for everything. ![Image](https://github.com/user-attachments/assets/8cd07744-6f4f-4fd5-8192-85bc6a3de860) ![Image](https://github.com/user-attachments/assets/c721544e-dfae-438f-85b4-6a36140f1115) ![Image](https://github.com/user-attachments/assets/102175e0-43bb-434d-9d7f-0bb6b9b488ff) ![Image](https://github.com/user-attachments/assets/c877f5b9-f987-4ec3-91da-5ac43285d1e3) ![Image](https://github.com/user-attachments/assets/df95efb3-ff2a-4078-bf05-7ca178e94a0d) ![Image](https://github.com/user-attachments/assets/18c69850-b254-483e-b0b9-ce259075d287)
yindo added the bug label 2026-02-16 00:17:49 -05:00
yindo closed this issue 2026-02-16 00:17:49 -05:00
yindo commented 2026-02-16 00:17:49 -05:00
Author
Owner
Copy Link
Copy Source

@hitoshyamamoto commented on GitHub (Feb 15, 2025):

I’m experiencing the same issue#621.

I’ve also been using LlamaParse frequently, and after upgrading to v0.6.1, I noticed that the new parameters (e.g., content_guideline_instruction) are not being applied properly when running through the Python script. Like you mentioned, it seems to ignore the guidelines.

I started noticing this issue about 12 hours ago, and since then, I haven’t been able to get consistent results.

Additionally, I came across this documentation from LlamaCloud, which mentions new parameters like user_prompt, system_prompt_append, and system_prompt. However, when checking the LlamaParse method in the API, I couldn’t find these parameters available. Do you know if they are planned to be added in a future release?
Also, I noticed that the parsing_instruction parameter is being referred to as "template" in JobParameters.

Hopefully, the devs can clarify this soon.

@hitoshyamamoto commented on GitHub (Feb 15, 2025): I’m experiencing the same [issue#621](https://github.com/run-llama/llama_cloud_services/issues/621). I’ve also been using LlamaParse frequently, and after upgrading to v0.6.1, I noticed that the new parameters (e.g., content_guideline_instruction) are not being applied properly when running through the Python script. Like you mentioned, it seems to ignore the guidelines. I started noticing this issue about 12 hours ago, and since then, I haven’t been able to get consistent results. Additionally, I came across [this documentation from LlamaCloud](https://docs.cloud.llamaindex.ai/llamaparse/features/prompts), which mentions new parameters like user_prompt, system_prompt_append, and system_prompt. However, when checking the LlamaParse method in the API, I couldn’t find these parameters available. Do you know if they are planned to be added in a future release? Also, I noticed that the parsing_instruction parameter is being referred to as "template" in JobParameters. Hopefully, the devs can clarify this soon.
yindo commented 2026-02-16 00:17:49 -05:00
Author
Owner
Copy Link
Copy Source

@mehwishh247 commented on GitHub (Feb 16, 2025):

I hope so too. I noticed the issue around the same time when my script broke. I also noticed through web GUI how they changed the prompt names to user prompt, system prompt append, and system prompt, which are equivalent to what you wrote. I think there might be a new update coming with changed parameter names. And the new and old parameters are just not linking together. Possibly they have merged a PR into something that is affecting the process instead of pushing it as final update.

@mehwishh247 commented on GitHub (Feb 16, 2025): I hope so too. I noticed the issue around the same time when my script broke. I also noticed through web GUI how they changed the prompt names to user prompt, system prompt append, and system prompt, which are equivalent to what you wrote. I think there might be a new update coming with changed parameter names. And the new and old parameters are just not linking together. Possibly they have merged a PR into something that is affecting the process instead of pushing it as final update.
yindo commented 2026-02-16 00:17:49 -05:00
Author
Owner
Copy Link
Copy Source

@judithnat commented on GitHub (Feb 16, 2025):

Also having the same problem. Was working perfectly but when retested yesterday, instead of nicely structured text we get just text extraction.

Lots of the documents have not been undated.

Parsing instruction seems to have been replaced by content_guideline_instruction etc. and even that is deprecated in favour of prompts. But NONE OF THESE WORK. When viewing job parameters in the sandbox they are all empty

Very difficult to deal with when working on production code

@judithnat commented on GitHub (Feb 16, 2025): Also having the same problem. Was working perfectly but when retested yesterday, instead of nicely structured text we get just text extraction. Lots of the documents have not been undated. Parsing instruction seems to have been replaced by content_guideline_instruction etc. and even that is deprecated in favour of prompts. But NONE OF THESE WORK. When viewing job parameters in the sandbox they are all empty Very difficult to deal with when working on production code
yindo commented 2026-02-16 00:17:49 -05:00
Author
Owner
Copy Link
Copy Source

@hitoshyamamoto commented on GitHub (Feb 16, 2025):

I received a response from one of the repository contributors yesterday (issue#621), suggesting that I use the parsing_instruction parameter. I also replied to him today after testing again, and unfortunately, the issue still persists.

For now, I'll be using parsing_instruction instead of the other parameters introduced in versions 0.6.0 and 0.6.1 until this is clarified or fixed in a future update.

Additionally, I tested the new parameters that appear in the GUI—user_prompt, system_prompt_append, and system_prompt—but they do not seem to be available in the LlamaParse() method. I’ll be waiting for the next update and looking forward to the contributor’s response regarding this.

Let’s hope we get an official response soon regarding these changes and whether the new parameters are expected to work properly.

@hitoshyamamoto commented on GitHub (Feb 16, 2025): I received a response from one of the repository contributors yesterday ([issue#621](https://github.com/run-llama/llama_cloud_services/issues/621)), suggesting that I use the **parsing_instruction** parameter. I also replied to him today after testing again, and unfortunately, the issue still persists. For now, I'll be using **parsing_instruction** instead of the other parameters introduced in versions 0.6.0 and 0.6.1 until this is clarified or fixed in a future update. Additionally, I tested the new parameters that appear in the GUI—user_prompt, system_prompt_append, and system_prompt—but they do not seem to be available in the LlamaParse() method. I’ll be waiting for the next update and looking forward to the contributor’s response regarding this. Let’s hope we get an official response soon regarding these changes and whether the new parameters are expected to work properly.
yindo commented 2026-02-16 00:17:49 -05:00
Author
Owner
Copy Link
Copy Source

@SaiCharan878 commented on GitHub (Jun 30, 2025):

Hi, I am having the same issue with all three prompts (user_prompt, system_prompt_append, and system_prompt). does anyone in the thread found any solution? the user_prompt is working fine in GUI but not in my script.

@SaiCharan878 commented on GitHub (Jun 30, 2025): Hi, I am having the same issue with all three prompts (user_prompt, system_prompt_append, and system_prompt). does anyone in the thread found any solution? the user_prompt is working fine in GUI but not in my script.
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
LlamaCloud
LlamaExtract
LlamaParse
bug
client
complex layout
documentation
duplicate
enhancement
figure
font
header footer
help wanted
ocr
parser
pdf_debug
pdf_debug
pull-request
Mirrored from GitHub Pull Request
 python-lib
question
server_issue
server_issue
ui
No Label bug
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
yindo
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: run-llama/llama_cloud_services#431
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?