chore: remove comment

feat: experimental package + json query engine (#613 )
feat: Add start in VSCode option and support python for dev container (#619 )
2026-07-14 20:48:32 -04:00 · 2024-03-08 15:51:34 -03:00 · 2024-03-07 14:34:55 -03:00 · 2024-03-07 17:19:08 +07:00 · 2024-03-06 23:41:42 -06:00 · 2024-03-06 23:32:02 -06:00
159 changed files with 2256 additions and 489 deletions
@@ -0,0 +1,5 @@
+---
+"llamaindex": patch
+---
+
+feat: experimental package + json query engine
@@ -0,0 +1,12 @@
+---
+"llamaindex": patch
+"@llamaindex/core-test": patch
+---
+
+- Add missing exports:
+  - `IndexStructType`,
+  - `IndexDict`,
+  - `jsonToIndexStruct`,
+  - `IndexList`,
+  - `IndexStruct`
+- Fix `IndexDict.toJson()` method
@@ -0,0 +1,5 @@
+---
+"create-llama": patch
+---
+
+Add "Start in VSCode" option to postInstallAction
@@ -0,0 +1,5 @@
+---
+"llamaindex": patch
+---
+
+Add streaming to agents
@@ -0,0 +1,5 @@
+---
+"create-llama": patch
+---
+
+Add devcontainers to generated code
@@ -0,0 +1,5 @@
+---
+"llamaindex": minor
+---
+
+Use parameter object for retrieve function of Retriever (to align usage with query function of QueryEngine)
@@ -11,5 +11,13 @@ module.exports = {
    "max-params": ["error", 4],
    "prefer-const": "error",
  },
+  overrides: [
+    {
+      files: ["examples/**/*.ts"],
+      rules: {
+        "turbo/no-undeclared-env-vars": "off",
+      },
+    },
+  ],
  ignorePatterns: ["dist/", "lib/"],
 };
@@ -0,0 +1,28 @@
+name: Publish
+on:
+  push:
+    branches:
+      - main
+
+jobs:
+  publish:
+    runs-on: ubuntu-latest
+
+    permissions:
+      contents: read
+      id-token: write
+
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Publish @llamaindex/env
+        run: npx jsr publish
+        working-directory: packages/env
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+
+      - name: Publish @llamaindex/core
+        run: npx jsr publish --allow-slow-types
+        working-directory: packages/core
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
@@ -44,6 +44,7 @@ test-results/
 playwright-report/
 blob-report/
 playwright/.cache/
+.tsbuildinfo

 # intellij
 **/.idea
@@ -1,5 +1,12 @@
 # docs

+## 0.0.4
+
+### Patch Changes
+
+- Updated dependencies [5116ad8]
+  - @llamaindex/env@0.0.5
+
 ## 0.0.3

 ### Patch Changes
@@ -23,3 +23,15 @@ const results = await queryEngine.query({
  query,
 });
 ```
+
+Per default, `HuggingFaceEmbedding` is using the `Xenova/all-MiniLM-L6-v2` model. You can change the model by passing the `modelType` parameter to the constructor.
+If you're not using a quantized model, set the `quantized` parameter to `false`.
+
+For example, to use the not quantized `BAAI/bge-small-en-v1.5` model, you can use the following code:
+
+```
+const embedModel = new HuggingFaceEmbedding({
+  modelType: "BAAI/bge-small-en-v1.5",
+  quantized: false,
+});
+```
@@ -53,10 +53,6 @@ const evaluator = new CorrectnessEvaluator({
  serviceContext: ctx,
 });

-const response = await queryEngine.query({
-  query,
-});
-
 const result = await evaluator.evaluateResponse({
  query,
  response,
@@ -100,7 +100,7 @@ const response = await queryEngine.query("<user_query>");
 ```ts
 import { SimilarityPostprocessor } from "llamaindex";

-nodes = await index.asRetriever().retrieve("test query str");
+nodes = await index.asRetriever().retrieve({ query: "test query str" });

 const processor = new SimilarityPostprocessor({
  similarityCutoff: 0.7,
@@ -11,7 +11,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Fetch nodes!
-const nodesWithScore = await retriever.retrieve("query string");
+const nodesWithScore = await retriever.retrieve({ query: "query string" });
 ```

 ## API Reference
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // جلب العقد!
-const nodesWithScore = await retriever.retrieve("سلسلة الاستعلام");
+const nodesWithScore = await retriever.retrieve({ query: "سلسلة الاستعلام" });
 ```

 ## مرجع الواجهة البرمجية (API Reference)
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Извличане на върхове!
-const nodesWithScore = await retriever.retrieve("query string");
+const nodesWithScore = await retriever.retrieve({ query: "query string" });
 ```

 ## API Reference (API справка)
@@ -13,7 +13,7 @@ const recuperador = vector_index.asRetriever();
 recuperador.similarityTopK = 3;

 // Obteniu els nodes!
-const nodesAmbPuntuació = await recuperador.retrieve("cadena de consulta");
+const nodesAmbPuntuació = await recuperador.retrieve({ query: "cadena de consulta" });
 ```

 ## Referència de l'API
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Získání uzlů!
-const nodesWithScore = await retriever.retrieve("dotazovací řetězec");
+const nodesWithScore = await retriever.retrieve({ query: "dotazovací řetězec" });
 ```

 ## API Reference (Odkazy na rozhraní)
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Hent noder!
-const nodesWithScore = await retriever.retrieve("forespørgselsstreng");
+const nodesWithScore = await retriever.retrieve({ query: "forespørgselsstreng" });
 ```

 ## API Reference
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Knoten abrufen!
-const nodesWithScore = await retriever.retrieve("Abfragezeichenfolge");
+const nodesWithScore = await retriever.retrieve({ query: "Abfragezeichenfolge" });
 ```

 ## API-Referenz
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Ανάκτηση κόμβων!
-const nodesWithScore = await retriever.retrieve("συμβολοσειρά ερωτήματος");
+const nodesWithScore = await retriever.retrieve({ query: "συμβολοσειρά ερωτήματος" });
 ```

 ## Αναφορά API
@@ -13,7 +13,7 @@ const recuperador = vector_index.asRetriever();
 recuperador.similarityTopK = 3;

 // ¡Obtener nodos!
-const nodosConPuntuación = await recuperador.retrieve("cadena de consulta");
+const nodosConPuntuación = await recuperador.retrieve({ query: "cadena de consulta" });
 ```

 ## Referencia de la API
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Too sõlmed!
-const nodesWithScore = await retriever.retrieve("päringu string");
+const nodesWithScore = await retriever.retrieve({ query: "päringu string" });
 ```

 ## API viide
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // بازیابی گره ها!
-const nodesWithScore = await retriever.retrieve("رشته پرس و جو");
+const nodesWithScore = await retriever.retrieve({ query: "رشته پرس و جو" });
 ```

 ## مرجع API
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Hae solmut!
-const nodesWithScore = await retriever.retrieve("kyselymerkkijono");
+const nodesWithScore = await retriever.retrieve({ query: "kyselymerkkijono" });
 ```

 ## API-viite
@@ -11,7 +11,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Récupérer les nœuds !
-const nodesWithScore = await retriever.retrieve("chaîne de requête");
+const nodesWithScore = await retriever.retrieve({ query: "chaîne de requête" });
 ```

 ## Référence de l'API
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // אחזור צמתים!
-const nodesWithScore = await retriever.retrieve("מחרוזת שאילתה");
+const nodesWithScore = await retriever.retrieve({ query: "מחרוזת שאילתה" });
 ```

 ## מדריך לממשק API
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // नोड्स प्राप्त करें!
-const nodesWithScore = await retriever.retrieve("क्वेरी स्ट्रिंग");
+const nodesWithScore = await retriever.retrieve({ query: "क्वेरी स्ट्रिंग" });
 ```

 ## एपीआई संदर्भ (API Reference)
@@ -13,7 +13,7 @@ const dohvatnik = vector_index.asRetriever();
 dohvatnik.similarityTopK = 3;

 // Dohvati čvorove!
-const čvoroviSaRezultatom = await dohvatnik.retrieve("upitni niz");
+const čvoroviSaRezultatom = await dohvatnik.retrieve({ query: "upitni niz" });
 ```

 ## API Referenca
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Node-ok lekérése!
-const nodesWithScore = await retriever.retrieve("lekérdezési karakterlánc");
+const nodesWithScore = await retriever.retrieve({ query: "lekérdezési karakterlánc" });
 ```

 ## API Referencia
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Mengambil node!
-const nodesWithScore = await retriever.retrieve("string query");
+const nodesWithScore = await retriever.retrieve({ query: "string query" });
 ```

 ## Referensi API
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Recupera i nodi!
-const nodesWithScore = await retriever.retrieve("stringa di query");
+const nodesWithScore = await retriever.retrieve({ query: "stringa di query" });
 ```

 ## Riferimento API
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // ノードを取得します！
-const nodesWithScore = await retriever.retrieve("クエリ文字列");
+const nodesWithScore = await retriever.retrieve({ query: "クエリ文字列" });
 ```

 ## API リファレンス
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // 노드를 가져옵니다!
-const nodesWithScore = await retriever.retrieve("쿼리 문자열");
+const nodesWithScore = await retriever.retrieve({ query: "쿼리 문자열" });
 ```

 ## API 참조
@@ -13,7 +13,7 @@ const gavėjas = vector_index.asRetriever();
 gavėjas.similarityTopK = 3;

 // Išgaunami mazgai!
-const mazgaiSuRezultatu = await gavėjas.retrieve("užklausos eilutė");
+const mazgaiSuRezultatu = await gavėjas.retrieve({ query: "užklausos eilutė" });
 ```

 ## API nuorodos (API Reference)
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Haal knooppunten op!
-const nodesWithScore = await retriever.retrieve("zoekopdracht");
+const nodesWithScore = await retriever.retrieve({ query: "zoekopdracht" });
 ```

 ## API Referentie
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Hent noder!
-const nodesWithScore = await retriever.retrieve("spørringsstreng");
+const nodesWithScore = await retriever.retrieve({ query: "spørringsstreng" });
 ```

 ## API-referanse
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Pobierz węzły!
-const nodesWithScore = await retriever.retrieve("ciąg zapytania");
+const nodesWithScore = await retriever.retrieve({ query: "ciąg zapytania" });
 ```

 ## Dokumentacja interfejsu API
@@ -13,7 +13,7 @@ const recuperador = vector_index.asRetriever();
 recuperador.similarityTopK = 3;

 // Buscar nós!
-const nósComPontuação = await recuperador.retrieve("string de consulta");
+const nósComPontuação = await recuperador.retrieve({ query: "string de consulta" });
 ```

 ## Referência da API
@@ -13,7 +13,7 @@ const recuperator = vector_index.asRetriever();
 recuperator.similarityTopK = 3;

 // Preia nodurile!
-const noduriCuScor = await recuperator.retrieve("șir de interogare");
+const noduriCuScor = await recuperator.retrieve({ query: "șir de interogare" });
 ```

 ## Referință API
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Получение узлов!
-const nodesWithScore = await retriever.retrieve("строка запроса");
+const nodesWithScore = await retriever.retrieve({ query: "строка запроса" });
 ```

 ## Справочник по API
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Dohvati čvorove!
-const nodesWithScore = await retriever.retrieve("upitni niz");
+const nodesWithScore = await retriever.retrieve({ query: "upitni niz" });
 ```

 ## API Referenca
@@ -13,7 +13,7 @@ const pridobitelj = vector_index.asRetriever();
 pridobitelj.similarityTopK = 3;

 // Pridobivanje vozlišč!
-const vozliščaZRezultatom = await pridobitelj.retrieve("poizvedbeni niz");
+const vozliščaZRezultatom = await pridobitelj.retrieve({ query: "poizvedbeni niz" });
 ```

 ## API Sklic
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Získajte uzly!
-const nodesWithScore = await retriever.retrieve("reťazec dotazu");
+const nodesWithScore = await retriever.retrieve({ query: "reťazec dotazu" });
 ```

 ## API Referencia
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Hämta noder!
-const nodesWithScore = await retriever.retrieve("frågesträng");
+const nodesWithScore = await retriever.retrieve({ query: "frågesträng" });
 ```

 ## API-referens
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // เรียกคืนโหนด!
-const nodesWithScore = await retriever.retrieve("query string");
+const nodesWithScore = await retriever.retrieve({ query: "query string" });
 ```

 ## API Reference (การอ้างอิง API)
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Düğümleri getir!
-const nodesWithScore = await retriever.retrieve("sorgu dizesi");
+const nodesWithScore = await retriever.retrieve({ query: "sorgu dizesi" });
 ```

 ## API Referansı
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Отримати вузли!
-const nodesWithScore = await retriever.retrieve("рядок запиту");
+const nodesWithScore = await retriever.retrieve({ query: "рядок запиту" });
 ```

 ## Довідник API
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // Lấy các node!
-const nodesWithScore = await retriever.retrieve("chuỗi truy vấn");
+const nodesWithScore = await retriever.retrieve({ query: "chuỗi truy vấn" });
 ```

 ## Tài liệu tham khảo API
@@ -11,7 +11,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // 获取节点！
-const nodesWithScore = await retriever.retrieve("查询字符串");
+const nodesWithScore = await retriever.retrieve({ query: "查询字符串" });
 ```

 ## API 参考
@@ -13,7 +13,7 @@ const retriever = vector_index.asRetriever();
 retriever.similarityTopK = 3;

 // 提取節點！
-const nodesWithScore = await retriever.retrieve("查詢字串");
+const nodesWithScore = await retriever.retrieve({ query: "查詢字串" });
 ```

 ## API 參考
@@ -1,6 +1,6 @@
 {
  "name": "docs",
-  "version": "0.0.3",
+  "version": "0.0.4",
  "private": true,
  "scripts": {
    "docusaurus": "docusaurus",
@@ -8,7 +8,7 @@ import {
 async function main() {
  // Load the documents
  const documents = await new SimpleDirectoryReader().loadData({
-    directoryPath: "node_modules/llamaindex/examples/",
+    directoryPath: "node_modules/llamaindex/examples",
  });

  // Create a vector index from the documents
@@ -0,0 +1,95 @@
+import { FunctionTool, OpenAIAgent } from "llamaindex";
+
+// Define a function to sum two numbers
+function sumNumbers({ a, b }: { a: number; b: number }): number {
+  return a + b;
+}
+
+// Define a function to divide two numbers
+function divideNumbers({ a, b }: { a: number; b: number }): number {
+  return a / b;
+}
+
+// Define the parameters of the sum function as a JSON schema
+const sumJSON = {
+  type: "object",
+  properties: {
+    a: {
+      type: "number",
+      description: "The first number",
+    },
+    b: {
+      type: "number",
+      description: "The second number",
+    },
+  },
+  required: ["a", "b"],
+};
+
+const divideJSON = {
+  type: "object",
+  properties: {
+    a: {
+      type: "number",
+      description: "The dividend a to divide",
+    },
+    b: {
+      type: "number",
+      description: "The divisor b to divide by",
+    },
+  },
+  required: ["a", "b"],
+};
+
+async function main() {
+  // Create a function tool from the sum function
+  const functionTool = new FunctionTool(sumNumbers, {
+    name: "sumNumbers",
+    description: "Use this function to sum two numbers",
+    parameters: sumJSON,
+  });
+
+  // Create a function tool from the divide function
+  const functionTool2 = new FunctionTool(divideNumbers, {
+    name: "divideNumbers",
+    description: "Use this function to divide two numbers",
+    parameters: divideJSON,
+  });
+
+  // Create an OpenAIAgent with the function tools
+  const agent = new OpenAIAgent({
+    tools: [functionTool, functionTool2],
+    verbose: true,
+  });
+
+  // Create a task to sum and divide numbers
+  const task = agent.createTask("How much is 5 + 5? then divide by 2");
+
+  let count = 0;
+
+  while (true) {
+    const stepOutput = await agent.runStep(task.taskId);
+
+    console.log(`Runnning step ${count++}`);
+    console.log(`======== OUTPUT ==========`);
+    if (stepOutput.output.response) {
+      console.log(stepOutput.output.response);
+    } else {
+      console.log(stepOutput.output.sources);
+    }
+    console.log(`==========================`);
+
+    if (stepOutput.isLast) {
+      const finalResponse = await agent.finalizeResponse(
+        task.taskId,
+        stepOutput,
+      );
+      console.log({ finalResponse });
+      break;
+    }
+  }
+}
+
+main().then(() => {
+  console.log("Done");
+});
@@ -8,7 +8,7 @@ import {
 async function main() {
  // Load the documents
  const documents = await new SimpleDirectoryReader().loadData({
-    directoryPath: "node_modules/llamaindex/examples/",
+    directoryPath: "node_modules/llamaindex/examples",
  });

  // Create a vector index from the documents
@@ -32,13 +32,31 @@ async function main() {
    verbose: true,
  });

-  // Chat with the agent
-  const response = await agent.chat({
-    message: "What was his salary?",
-  });
+  const task = agent.createTask("What was his salary?");

-  // Print the response
-  console.log(String(response));
+  let count = 0;
+
+  while (true) {
+    const stepOutput = await agent.runStep(task.taskId);
+
+    console.log(`Runnning step ${count++}`);
+    console.log(`======== OUTPUT ==========`);
+    if (stepOutput.output.response) {
+      console.log(stepOutput.output.response);
+    } else {
+      console.log(stepOutput.output.sources);
+    }
+    console.log(`==========================`);
+
+    if (stepOutput.isLast) {
+      const finalResponse = await agent.finalizeResponse(
+        task.taskId,
+        stepOutput,
+      );
+      console.log({ finalResponse });
+      break;
+    }
+  }
 }

 main().then(() => {
@@ -0,0 +1,90 @@
+import { FunctionTool, ReActAgent } from "llamaindex";
+
+// Define a function to sum two numbers
+function sumNumbers({ a, b }: { a: number; b: number }): number {
+  return a + b;
+}
+
+// Define a function to divide two numbers
+function divideNumbers({ a, b }: { a: number; b: number }): number {
+  return a / b;
+}
+
+// Define the parameters of the sum function as a JSON schema
+const sumJSON = {
+  type: "object",
+  properties: {
+    a: {
+      type: "number",
+      description: "The first number",
+    },
+    b: {
+      type: "number",
+      description: "The second number",
+    },
+  },
+  required: ["a", "b"],
+};
+
+const divideJSON = {
+  type: "object",
+  properties: {
+    a: {
+      type: "number",
+      description: "The dividend",
+    },
+    b: {
+      type: "number",
+      description: "The divisor",
+    },
+  },
+  required: ["a", "b"],
+};
+
+async function main() {
+  // Create a function tool from the sum function
+  const functionTool = new FunctionTool(sumNumbers, {
+    name: "sumNumbers",
+    description: "Use this function to sum two numbers",
+    parameters: sumJSON,
+  });
+
+  // Create a function tool from the divide function
+  const functionTool2 = new FunctionTool(divideNumbers, {
+    name: "divideNumbers",
+    description: "Use this function to divide two numbers",
+    parameters: divideJSON,
+  });
+
+  // Create an OpenAIAgent with the function tools
+  const agent = new ReActAgent({
+    tools: [functionTool, functionTool2],
+    verbose: true,
+  });
+
+  const task = agent.createTask("Divide 16 by 2 then add 20");
+
+  let count = 0;
+
+  while (true) {
+    const stepOutput = await agent.runStep(task.taskId);
+
+    console.log(`Runnning step ${count++}`);
+    console.log(`======== OUTPUT ==========`);
+    console.log(stepOutput.output);
+    console.log(`==========================`);
+
+    if (stepOutput.isLast) {
+      const finalResponse = await agent.finalizeResponse(
+        task.taskId,
+        stepOutput,
+      );
+      console.log({ finalResponse });
+      break;
+    }
+  }
+}
+
+main().then(() => {
+  console.log("Done");
+});
@@ -0,0 +1,77 @@
+import { FunctionTool, OpenAIAgent } from "llamaindex";
+
+// Define a function to sum two numbers
+function sumNumbers({ a, b }: { a: number; b: number }): number {
+  return a + b;
+}
+
+// Define a function to divide two numbers
+function divideNumbers({ a, b }: { a: number; b: number }): number {
+  return a / b;
+}
+
+// Define the parameters of the sum function as a JSON schema
+const sumJSON = {
+  type: "object",
+  properties: {
+    a: {
+      type: "number",
+      description: "The first number",
+    },
+    b: {
+      type: "number",
+      description: "The second number",
+    },
+  },
+  required: ["a", "b"],
+};
+
+const divideJSON = {
+  type: "object",
+  properties: {
+    a: {
+      type: "number",
+      description: "The dividend",
+    },
+    b: {
+      type: "number",
+      description: "The divisor",
+    },
+  },
+  required: ["a", "b"],
+};
+
+async function main() {
+  // Create a function tool from the sum function
+  const functionTool = new FunctionTool(sumNumbers, {
+    name: "sumNumbers",
+    description: "Use this function to sum two numbers",
+    parameters: sumJSON,
+  });
+
+  // Create a function tool from the divide function
+  const functionTool2 = new FunctionTool(divideNumbers, {
+    name: "divideNumbers",
+    description: "Use this function to divide two numbers",
+    parameters: divideJSON,
+  });
+
+  // Create an OpenAIAgent with the function tools
+  const agent = new OpenAIAgent({
+    tools: [functionTool, functionTool2],
+    verbose: false,
+  });
+
+  const stream = await agent.chat({
+    message: "Divide 16 by 2 then add 20",
+    stream: true,
+  });
+
+  for await (const chunk of stream.response) {
+    process.stdout.write(chunk.response);
+  }
+}
+
+main().then(() => {
+  console.log("\nDone");
+});
@@ -3,6 +3,7 @@ import { Anthropic } from "llamaindex";
 (async () => {
  const anthropic = new Anthropic({
    apiKey: process.env.ANTHROPIC_API_KEY,
+    model: "claude-3-opus",
  });
  const result = await anthropic.chat({
    messages: [
@@ -0,0 +1,34 @@
+import { Anthropic, SimpleChatEngine, SimpleChatHistory } from "llamaindex";
+import { stdin as input, stdout as output } from "node:process";
+import readline from "node:readline/promises";
+
+(async () => {
+  const llm = new Anthropic({
+    apiKey: process.env.ANTHROPIC_API_KEY,
+    model: "claude-3-opus",
+  });
+  // chatHistory will store all the messages in the conversation
+  const chatHistory = new SimpleChatHistory({
+    messages: [
+      {
+        content: "You want to talk in rhymes.",
+        role: "system",
+      },
+    ],
+  });
+  const chatEngine = new SimpleChatEngine({
+    llm,
+    chatHistory,
+  });
+  const rl = readline.createInterface({ input, output });
+
+  while (true) {
+    const query = await rl.question("User: ");
+    process.stdout.write("Assistant: ");
+    const stream = await chatEngine.chat({ message: query, stream: true });
+    for await (const chunk of stream) {
+      process.stdout.write(chunk.response);
+    }
+    process.stdout.write("\n");
+  }
+})();
@@ -0,0 +1,23 @@
+import { Anthropic } from "llamaindex";
+
+(async () => {
+  const anthropic = new Anthropic({
+    apiKey: process.env.ANTHROPIC_API_KEY,
+    model: "claude-instant-1.2",
+  });
+  const stream = await anthropic.chat({
+    messages: [
+      { content: "You want to talk in rhymes.", role: "system" },
+      {
+        content:
+          "How much wood would a woodchuck chuck if a woodchuck could chuck wood?",
+        role: "user",
+      },
+    ],
+    stream: true,
+  });
+
+  for await (const chunk of stream) {
+    process.stdout.write(chunk.delta);
+  }
+})();
@@ -27,9 +27,9 @@ async function main() {
  // retrieve documents using the index
  const index = await createIndex();
  const retriever = index.asRetriever({ similarityTopK: 3 });
-  const results = await retriever.retrieve(
-    "what are Vincent van Gogh's famous paintings",
-  );
+  const results = await retriever.retrieve({
+    query: "what are Vincent van Gogh's famous paintings",
+  });
  for (const result of results) {
    const node = result.node;
    if (!node) {
@@ -1,4 +1,4 @@
-import { OpenAI } from "llamaindex";
+import { OpenAI, OpenAIEmbedding } from "llamaindex";

 (async () => {
  const llm = new OpenAI({ model: "gpt-4-1106-preview", temperature: 0.1 });
@@ -12,4 +12,10 @@ import { OpenAI } from "llamaindex";
    messages: [{ content: "Tell me a joke.", role: "user" }],
  });
  console.log(response2.message.content);
+
+  // embeddings
+  const embedModel = new OpenAIEmbedding();
+  const texts = ["hello", "world"];
+  const embeddings = await embedModel.getTextEmbeddingsBatch(texts);
+  console.log(`\nWe have ${embeddings.length} embeddings`);
 })();
@@ -7,8 +7,9 @@ There are two scripts available here: load-docs.ts and query.ts
 You'll need a Pinecone account, project, and index. Pinecone does not allow automatic creation of indexes on the free plan,
 so this vector store does not check and create the index (unlike, e.g., the PGVectorStore)

-Set the **PINECONE_API_KEY** and **PINECONE_ENVIRONMENT** environment variables to match your specific values. You will likely also need to set **PINECONE_INDEX_NAME**, unless your
-index is the default value "llama".
+Set the **PINECONE_API_KEY** and **PINECONE_ENVIRONMENT** environment variables to match your specific values.
+You will likely also need to set **PINECONE_INDEX_NAME**, unless your index is the default value "llama".
+By default, all operations take place inside the default namespace '', but you can set **PINECONE_NAMESPACE** to a different value if you need to.

 You'll also need a value for OPENAI_API_KEY in your environment.

@@ -13,8 +13,8 @@
    "type-check": "tsc -b --diagnostics",
    "release": "pnpm run build:release && changeset publish",
    "new-llamaindex": "pnpm run build:release && changeset version --ignore create-llama",
-    "new-create-llama": "pnpm run build:release && changeset version --ignore llamaindex",
-    "new-snapshots": "pnpm run build:release && changeset version --snapshot"
+    "new-create-llama": "pnpm run build:release && changeset version --ignore llamaindex --ignore @llamaindex/core-test",
+    "new-experimental": "pnpm run build:release && changeset version --ignore create-llama"
  },
  "devDependencies": {
    "@changesets/cli": "^2.27.1",
@@ -1 +1,3 @@
 .turbo
+README.md
+LICENSE
@@ -1,5 +1,29 @@
 # llamaindex

+## 0.1.21
+
+### Patch Changes
+
+- 552a61a: Add quantized parameter to HuggingFaceEmbedding
+- d824876: Add support for Claude 3
+
+## 0.1.20
+
+### Patch Changes
+
+- 64683a5: fix: prefix messages always true
+- 698cd9c: fix: step wise agent + examples
+- 7257751: fixed removeRefDocNode and persist store on delete
+- 5116ad8: fix: compatibility issue with Deno
+- Updated dependencies [5116ad8]
+  - @llamaindex/env@0.0.5
+
+## 0.1.19
+
+### Patch Changes
+
+- 026d068: feat: enhance pinecone usage
+
 ## 0.1.18

 ### Patch Changes
@@ -0,0 +1,8 @@
+{
+  "name": "@llamaindex/core",
+  "version": "0.1.21",
+  "exports": "./src/index.ts",
+  "imports": {
+    "@llamaindex/env": "jsr:@llamaindex/env@0.0.5"
+  }
+}
@@ -1,22 +1,22 @@
 {
  "name": "llamaindex",
-  "version": "0.1.18",
+  "version": "0.1.21",
  "license": "MIT",
  "type": "module",
  "dependencies": {
-    "@anthropic-ai/sdk": "^0.13.0",
+    "@anthropic-ai/sdk": "^0.15.0",
    "@aws-crypto/sha256-js": "^5.2.0",
    "@datastax/astra-db-ts": "^0.1.4",
-    "@types/lodash": "^4.14.202",
-    "@types/node": "^18.19.14",
-    "@types/papaparse": "^5.3.14",
-    "@types/pg": "^8.11.0",
-    "@llamaindex/cloud": "^0.0.1",
+    "@llamaindex/cloud": "0.0.4",
    "@llamaindex/env": "workspace:*",
    "@mistralai/mistralai": "^0.0.10",
    "@notionhq/client": "^2.2.14",
    "@pinecone-database/pinecone": "^2.0.1",
    "@qdrant/js-client-rest": "^1.7.0",
+    "@types/lodash": "^4.14.202",
+    "@types/node": "^18.19.14",
+    "@types/papaparse": "^5.3.14",
+    "@types/pg": "^8.11.0",
    "@xenova/transformers": "^2.15.0",
    "assemblyai": "^4.2.2",
    "chromadb": "~1.7.3",
@@ -92,10 +92,11 @@
  "scripts": {
    "lint": "eslint .",
    "build": "rm -rf ./dist && pnpm run build:esm && pnpm run build:cjs && pnpm run build:type",
-    "build:esm": "swc src -d dist --strip-leading-paths --config-file .swcrc",
-    "build:cjs": "swc src -d dist/cjs --strip-leading-paths --config-file .cjs.swcrc",
-    "build:type": "pnpm run -w type-check",
-    "postbuild": "node -e \"require('fs').writeFileSync('./dist/cjs/package.json', JSON.stringify({ type: 'commonjs' }))\"",
+    "build:esm": "swc src -d dist --strip-leading-paths --config-file ../../.swcrc",
+    "build:cjs": "swc src -d dist/cjs --strip-leading-paths --config-file ../../.cjs.swcrc",
+    "build:type": "tsc -p tsconfig.json",
+    "copy": "cp -r ../../README.md ../../LICENSE .",
+    "postbuild": "pnpm run copy && node -e \"require('fs').writeFileSync('./dist/cjs/package.json', JSON.stringify({ type: 'commonjs' }))\"",
    "circular-check": "madge -c ./src/index.ts",
    "dev": "concurrently \"pnpm run build:esm --watch\" \"pnpm run build:cjs --watch\" \"pnpm run build:type --watch\""
  }
@@ -2,14 +2,16 @@ import type { Event } from "./callbacks/CallbackManager.js";
 import type { NodeWithScore } from "./Node.js";
 import type { ServiceContext } from "./ServiceContext.js";

+export type RetrieveParams = {
+  query: string;
+  parentEvent?: Event;
+  preFilters?: unknown;
+};
+
 /**
 * Retrievers retrieve the nodes that most closely match our query in similarity.
 */
 export interface BaseRetriever {
-  retrieve(
-    query: string,
-    parentEvent?: Event,
-    preFilters?: unknown,
-  ): Promise<NodeWithScore[]>;
+  retrieve(params: RetrieveParams): Promise<NodeWithScore[]>;
  getServiceContext(): ServiceContext;
 }
@@ -37,8 +37,6 @@ export class OpenAIAgent extends AgentRunner {
    toolRetriever,
    systemPrompt,
  }: OpenAIAgentParams) {
-    prefixMessages = prefixMessages || [];
-
    llm = llm ?? new OpenAI({ model: "gpt-3.5-turbo-0613" });

    if (systemPrompt) {
@@ -1,10 +1,10 @@
-// Assuming that the necessary interfaces and classes (like BaseTool, OpenAI, ChatMessage, CallbackManager, etc.) are defined elsewhere
-
 import { randomUUID } from "@llamaindex/env";
+import { Response } from "../../Response.js";
 import type { CallbackManager } from "../../callbacks/CallbackManager.js";
 import {
  AgentChatResponse,
  ChatResponseMode,
+  StreamingAgentChatResponse,
 } from "../../engines/chat/types.js";
 import type {
  ChatMessage,
@@ -12,6 +12,7 @@ import type {
  ChatResponseChunk,
 } from "../../llm/index.js";
 import { OpenAI } from "../../llm/index.js";
+import { streamConverter, streamReducer } from "../../llm/utils.js";
 import { ChatMemoryBuffer } from "../../memory/ChatMemoryBuffer.js";
 import type { ObjectRetriever } from "../../objects/base.js";
 import type { ToolOutput } from "../../tools/types.js";
@@ -192,13 +193,40 @@ export class OpenAIAgentWorker implements AgentWorker {
  private _processMessage(
    task: Task,
    chatResponse: ChatResponse,
-  ): AgentChatResponse | AsyncIterable<ChatResponseChunk> {
+  ): AgentChatResponse {
    const aiMessage = chatResponse.message;
    task.extraState.newMemory.put(aiMessage);

    return new AgentChatResponse(aiMessage.content, task.extraState.sources);
  }

+  private async _getStreamAiResponse(
+    task: Task,
+    llmChatKwargs: any,
+  ): Promise<StreamingAgentChatResponse> {
+    const stream = await this.llm.chat({
+      stream: true,
+      ...llmChatKwargs,
+    });
+
+    const iterator = streamConverter(
+      streamReducer({
+        stream,
+        initialValue: "",
+        reducer: (accumulator, part) => (accumulator += part.delta),
+        finished: (accumulator) => {
+          task.extraState.newMemory.put({
+            content: accumulator,
+            role: "assistant",
+          });
+        },
+      }),
+      (r: ChatResponseChunk) => new Response(r.delta),
+    );
+
+    return new StreamingAgentChatResponse(iterator, task.extraState.sources);
+  }
+
  /**
   * Get agent response.
   * @param task: task
@@ -210,7 +238,7 @@ export class OpenAIAgentWorker implements AgentWorker {
    task: Task,
    mode: ChatResponseMode,
    llmChatKwargs: any,
-  ): Promise<AgentChatResponse> {
+  ): Promise<AgentChatResponse | StreamingAgentChatResponse> {
    if (mode === ChatResponseMode.WAIT) {
      const chatResponse = (await this.llm.chat({
        stream: false,
@@ -218,9 +246,11 @@ export class OpenAIAgentWorker implements AgentWorker {
      })) as unknown as ChatResponse;

      return this._processMessage(task, chatResponse) as AgentChatResponse;
-    } else {
-      throw new Error("Not implemented");
+    } else if (mode === ChatResponseMode.STREAM) {
+      return this._getStreamAiResponse(task, llmChatKwargs);
    }
+
+    throw new Error("Invalid mode");
  }

  /**
@@ -4,6 +4,7 @@ import type { ChatEngineAgentParams } from "../../engines/chat/index.js";
 import {
  AgentChatResponse,
  ChatResponseMode,
+  StreamingAgentChatResponse,
 } from "../../engines/chat/index.js";
 import type { ChatMessage, LLM } from "../../llm/index.js";
 import { ChatMemoryBuffer } from "../../memory/ChatMemoryBuffer.js";
@@ -14,7 +15,7 @@ import { AgentState, BaseAgentRunner, TaskState } from "./types.js";

 const validateStepFromArgs = (
  taskId: string,
-  input: string,
+  input?: string | null,
  step?: any,
  kwargs?: any,
 ): TaskStep | undefined => {
@@ -24,6 +25,7 @@ const validateStepFromArgs = (
    }
    return step;
  } else {
+    if (!input) return;
    return new TaskStep(taskId, step, input, kwargs);
  }
 };
@@ -194,7 +196,7 @@ export class AgentRunner extends BaseAgentRunner {
   */
  async runStep(
    taskId: string,
-    input: string,
+    input?: string | null,
    step?: TaskStep,
    kwargs: any = {},
  ): Promise<TaskStepOutput> {
@@ -230,23 +232,26 @@ export class AgentRunner extends BaseAgentRunner {
    taskId: string,
    stepOutput: TaskStepOutput,
    kwargs?: any,
-  ): Promise<AgentChatResponse> {
+  ): Promise<AgentChatResponse | StreamingAgentChatResponse> {
    if (!stepOutput) {
      stepOutput =
        this.getCompletedSteps(taskId)[
          this.getCompletedSteps(taskId).length - 1
        ];
    }
+
    if (!stepOutput.isLast) {
      throw new Error(
        "finalizeResponse can only be called on the last step output",
      );
    }

-    if (!(stepOutput.output instanceof AgentChatResponse)) {
-      throw new Error(
-        `When \`isLast\` is True, cur_step_output.output must be AGENT_CHAT_RESPONSE_TYPE: ${stepOutput.output}`,
-      );
+    if (!(stepOutput.output instanceof StreamingAgentChatResponse)) {
+      if (!(stepOutput.output instanceof AgentChatResponse)) {
+        throw new Error(
+          `When \`isLast\` is True, cur_step_output.output must be AGENT_CHAT_RESPONSE_TYPE: ${stepOutput.output}`,
+        );
+      }
    }

    this.agentWorker.finalizeTask(this.getTask(taskId), kwargs);
@@ -261,20 +266,32 @@ export class AgentRunner extends BaseAgentRunner {
  protected async _chat({
    message,
    toolChoice,
-  }: ChatEngineAgentParams & { mode: ChatResponseMode }) {
+    stream,
+  }: ChatEngineAgentParams): Promise<AgentChatResponse>;
+  protected async _chat({
+    message,
+    toolChoice,
+    stream,
+  }: ChatEngineAgentParams & {
+    stream: true;
+  }): Promise<StreamingAgentChatResponse>;
+  protected async _chat({
+    message,
+    toolChoice,
+    stream,
+  }: ChatEngineAgentParams): Promise<
+    AgentChatResponse | StreamingAgentChatResponse
+  > {
    const task = this.createTask(message as string);

    let resultOutput;

+    const mode = stream ? ChatResponseMode.STREAM : ChatResponseMode.WAIT;
+
    while (true) {
-      const curStepOutput = await this._runStep(
-        task.taskId,
-        undefined,
-        ChatResponseMode.WAIT,
-        {
-          toolChoice,
-        },
-      );
+      const curStepOutput = await this._runStep(task.taskId, undefined, mode, {
+        toolChoice,
+      });

      if (curStepOutput.isLast) {
        resultOutput = curStepOutput;
@@ -298,7 +315,26 @@ export class AgentRunner extends BaseAgentRunner {
    message,
    chatHistory,
    toolChoice,
-  }: ChatEngineAgentParams): Promise<AgentChatResponse> {
+    stream,
+  }: ChatEngineAgentParams & {
+    stream?: false;
+  }): Promise<AgentChatResponse>;
+  public async chat({
+    message,
+    chatHistory,
+    toolChoice,
+    stream,
+  }: ChatEngineAgentParams & {
+    stream: true;
+  }): Promise<StreamingAgentChatResponse>;
+  public async chat({
+    message,
+    chatHistory,
+    toolChoice,
+    stream,
+  }: ChatEngineAgentParams): Promise<
+    AgentChatResponse | StreamingAgentChatResponse
+  > {
    if (!toolChoice) {
      toolChoice = this.defaultToolChoice;
    }
@@ -307,7 +343,7 @@ export class AgentRunner extends BaseAgentRunner {
      message,
      chatHistory,
      toolChoice,
-      mode: ChatResponseMode.WAIT,
+      stream,
    });

    return chatResponse;
@@ -1,4 +1,7 @@
-import type { AgentChatResponse } from "../../engines/chat/index.js";
+import type {
+  AgentChatResponse,
+  StreamingAgentChatResponse,
+} from "../../engines/chat/index.js";
 import type { Task, TaskStep, TaskStepOutput } from "../types.js";
 import { BaseAgent } from "../types.js";

@@ -57,7 +60,7 @@ export abstract class BaseAgentRunner extends BaseAgent {
    taskId: string,
    stepOutput: TaskStepOutput,
    kwargs?: any,
-  ): Promise<AgentChatResponse>;
+  ): Promise<AgentChatResponse | StreamingAgentChatResponse>;

  abstract undoStep(taskId: string): void;
 }
@@ -1,7 +1,9 @@
 import type {
  AgentChatResponse,
  ChatEngineAgentParams,
+  StreamingAgentChatResponse,
 } from "../engines/chat/index.js";
+
 import type { QueryEngineParamsNonStreaming } from "../types.js";

 export interface AgentWorker {
@@ -12,11 +14,15 @@ export interface AgentWorker {
 }

 interface BaseChatEngine {
-  chat(params: ChatEngineAgentParams): Promise<AgentChatResponse>;
+  chat(
+    params: ChatEngineAgentParams,
+  ): Promise<AgentChatResponse | StreamingAgentChatResponse>;
 }

 interface BaseQueryEngine {
-  query(params: QueryEngineParamsNonStreaming): Promise<AgentChatResponse>;
+  query(
+    params: QueryEngineParamsNonStreaming,
+  ): Promise<AgentChatResponse | StreamingAgentChatResponse>;
 }

 /**
@@ -31,7 +37,10 @@ export abstract class BaseAgent implements BaseChatEngine, BaseQueryEngine {
    return [];
  }

-  abstract chat(params: ChatEngineAgentParams): Promise<AgentChatResponse>;
+  abstract chat(
+    params: ChatEngineAgentParams,
+  ): Promise<AgentChatResponse | StreamingAgentChatResponse>;
+
  abstract reset(): void;

  /**
@@ -41,7 +50,7 @@ export abstract class BaseAgent implements BaseChatEngine, BaseQueryEngine {
   */
  async query(
    params: QueryEngineParamsNonStreaming,
-  ): Promise<AgentChatResponse> {
+  ): Promise<AgentChatResponse | StreamingAgentChatResponse> {
    // Handle non-streaming query
    const agentResponse = await this.chat({
      message: params.query,
@@ -161,13 +170,13 @@ export class TaskStep implements ITaskStep {
 * @param isLast: isLast
 */
 export class TaskStepOutput {
-  output: unknown;
+  output: any;
  taskStep: TaskStep;
  nextSteps: TaskStep[];
  isLast: boolean;

  constructor(
-    output: unknown,
+    output: any,
    taskStep: TaskStep,
    nextSteps: TaskStep[],
    isLast: boolean = false,
@@ -3,7 +3,7 @@ import { RetrieverQueryEngine } from "../engines/query/RetrieverQueryEngine.js";
 import type { BaseNodePostprocessor } from "../postprocessors/types.js";
 import type { BaseSynthesizer } from "../synthesizers/types.js";
 import type { BaseQueryEngine } from "../types.js";
-import type { RetrieveParams } from "./LlamaCloudRetriever.js";
+import type { CloudRetrieveParams } from "./LlamaCloudRetriever.js";
 import { LlamaCloudRetriever } from "./LlamaCloudRetriever.js";
 import type { CloudConstructorParams } from "./types.js";

@@ -14,7 +14,7 @@ export class LlamaCloudIndex {
    this.params = params;
  }

-  asRetriever(params: RetrieveParams = {}): BaseRetriever {
+  asRetriever(params: CloudRetrieveParams = {}): BaseRetriever {
    return new LlamaCloudRetriever({ ...this.params, ...params });
  }

@@ -23,7 +23,7 @@ export class LlamaCloudIndex {
      responseSynthesizer?: BaseSynthesizer;
      preFilters?: unknown;
      nodePostprocessors?: BaseNodePostprocessor[];
-    } & RetrieveParams,
+    } & CloudRetrieveParams,
  ): BaseQueryEngine {
    const retriever = new LlamaCloudRetriever({
      ...this.params,
@@ -2,15 +2,14 @@ import type { PlatformApi, PlatformApiClient } from "@llamaindex/cloud";
 import { globalsHelper } from "../GlobalsHelper.js";
 import type { NodeWithScore } from "../Node.js";
 import { ObjectType, jsonToNode } from "../Node.js";
-import type { BaseRetriever } from "../Retriever.js";
+import type { BaseRetriever, RetrieveParams } from "../Retriever.js";
 import type { ServiceContext } from "../ServiceContext.js";
 import { serviceContextFromDefaults } from "../ServiceContext.js";
-import type { Event } from "../callbacks/CallbackManager.js";
 import type { ClientParams, CloudConstructorParams } from "./types.js";
 import { DEFAULT_PROJECT_NAME } from "./types.js";
 import { getClient } from "./utils.js";

-export type RetrieveParams = Omit<
+export type CloudRetrieveParams = Omit<
  PlatformApi.RetrievalParams,
  "query" | "searchFilters" | "pipelineId" | "className"
 > & { similarityTopK?: number };
@@ -18,7 +17,7 @@ export type RetrieveParams = Omit<
 export class LlamaCloudRetriever implements BaseRetriever {
  client?: PlatformApiClient;
  clientParams: ClientParams;
-  retrieveParams: RetrieveParams;
+  retrieveParams: CloudRetrieveParams;
  projectName: string = DEFAULT_PROJECT_NAME;
  pipelineName: string;
  serviceContext: ServiceContext;
@@ -35,7 +34,7 @@ export class LlamaCloudRetriever implements BaseRetriever {
    });
  }

-  constructor(params: CloudConstructorParams & RetrieveParams) {
+  constructor(params: CloudConstructorParams & CloudRetrieveParams) {
    this.clientParams = { apiKey: params.apiKey, baseUrl: params.baseUrl };
    if (params.similarityTopK) {
      params.denseSimilarityTopK = params.similarityTopK;
@@ -55,11 +54,11 @@ export class LlamaCloudRetriever implements BaseRetriever {
    return this.client;
  }

-  async retrieve(
-    query: string,
-    parentEvent?: Event | undefined,
-    preFilters?: unknown,
-  ): Promise<NodeWithScore[]> {
+  async retrieve({
+    query,
+    parentEvent,
+    preFilters,
+  }: RetrieveParams): Promise<NodeWithScore[]> {
    const pipelines = await (
      await this.getClient()
    ).pipeline.searchPipelines({
@@ -1,4 +1,5 @@
 import type { PlatformApiClient } from "@llamaindex/cloud";
+import { getEnv } from "@llamaindex/env";
 import type { ClientParams } from "./types.js";
 import { DEFAULT_BASE_URL } from "./types.js";

@@ -7,8 +8,8 @@ export async function getClient({
  baseUrl,
 }: ClientParams = {}): Promise<PlatformApiClient> {
  // Get the environment variables or use defaults
-  baseUrl = baseUrl ?? process.env.LLAMA_CLOUD_BASE_URL ?? DEFAULT_BASE_URL;
-  apiKey = apiKey ?? process.env.LLAMA_CLOUD_API_KEY;
+  baseUrl = baseUrl ?? getEnv("LLAMA_CLOUD_BASE_URL") ?? DEFAULT_BASE_URL;
+  apiKey = apiKey ?? getEnv("LLAMA_CLOUD_API_KEY");

  const { PlatformApiClient } = await import("@llamaindex/cloud");

@@ -20,6 +20,7 @@ export enum HuggingFaceEmbeddingModelType {
 */
 export class HuggingFaceEmbedding extends BaseEmbedding {
  modelType: string = HuggingFaceEmbeddingModelType.XENOVA_ALL_MINILM_L6_V2;
+  quantized: boolean = true;

  private extractor: any;

@@ -31,7 +32,9 @@ export class HuggingFaceEmbedding extends BaseEmbedding {
  async getExtractor() {
    if (!this.extractor) {
      const { pipeline } = await import("@xenova/transformers");
-      this.extractor = await pipeline("feature-extraction", this.modelType);
+      this.extractor = await pipeline("feature-extraction", this.modelType, {
+        quantized: this.quantized,
+      });
    }
    return this.extractor;
  }
@@ -1,9 +1,10 @@
+import { getEnv } from "@llamaindex/env";
 import { OpenAIEmbedding } from "./OpenAIEmbedding.js";

 export class FireworksEmbedding extends OpenAIEmbedding {
  constructor(init?: Partial<OpenAIEmbedding>) {
    const {
-      apiKey = process.env.FIREWORKS_API_KEY,
+      apiKey = getEnv("FIREWORKS_API_KEY"),
      additionalSessionOptions = {},
      model = "nomic-ai/nomic-embed-text-v1.5",
      ...rest
@@ -1,9 +1,10 @@
+import { getEnv } from "@llamaindex/env";
 import { OpenAIEmbedding } from "./OpenAIEmbedding.js";

 export class TogetherEmbedding extends OpenAIEmbedding {
  constructor(init?: Partial<OpenAIEmbedding>) {
    const {
-      apiKey = process.env.TOGETHER_API_KEY,
+      apiKey = getEnv("TOGETHER_API_KEY"),
      additionalSessionOptions = {},
      model = "togethercomputer/m2-bert-80M-32k-retrieval",
      ...rest
@@ -64,10 +64,10 @@ export class DefaultContextGenerator
        tags: ["final"],
      };
    }
-    const sourceNodesWithScore = await this.retriever.retrieve(
-      message,
+    const sourceNodesWithScore = await this.retriever.retrieve({
+      query: message,
      parentEvent,
-    );
+    });

    const nodes = await this.applyNodePostprocessors(
      sourceNodesWithScore,
@@ -27,6 +27,7 @@ export interface ChatEngineParamsNonStreaming extends ChatEngineParamsBase {

 export interface ChatEngineAgentParams extends ChatEngineParamsBase {
  toolChoice?: string | Record<string, any>;
+  stream?: boolean;
 }

 /**
@@ -86,3 +87,20 @@ export class AgentChatResponse {
    return this.response ?? "";
  }
 }
+
+export class StreamingAgentChatResponse {
+  response: AsyncIterable<Response>;
+
+  sources: ToolOutput[];
+  sourceNodes?: BaseNode[];
+
+  constructor(
+    response: AsyncIterable<Response>,
+    sources?: ToolOutput[],
+    sourceNodes?: BaseNode[],
+  ) {
+    this.response = response;
+    this.sources = sources ?? [];
+    this.sourceNodes = sourceNodes ?? [];
+  }
+}
@@ -63,11 +63,11 @@ export class RetrieverQueryEngine
  }

  private async retrieve(query: string, parentEvent: Event) {
-    const nodes = await this.retriever.retrieve(
+    const nodes = await this.retriever.retrieve({
      query,
      parentEvent,
-      this.preFilters,
-    );
+      preFilters: this.preFilters,
+    });

    return await this.applyNodePostprocessors(nodes, query);
  }
@@ -30,3 +30,4 @@ export * from "./selectors/index.js";
 export * from "./storage/index.js";
 export * from "./synthesizers/index.js";
 export * from "./tools/index.js";
+export * from "./types.js";
@@ -1,4 +1,6 @@
 export * from "./BaseIndex.js";
+export * from "./IndexStruct.js";
+export * from "./json-to-index-struct.js";
 export * from "./keyword/index.js";
 export * from "./summary/index.js";
 export * from "./vectorStore/index.js";
@@ -24,9 +24,15 @@ export class IndexDict extends IndexStruct {
  }

  toJson(): Record<string, unknown> {
+    const nodesDict: Record<string, unknown> = {};
+
+    for (const [key, node] of Object.entries(this.nodesDict)) {
+      nodesDict[key] = node.toJSON();
+    }
+
    return {
      ...super.toJson(),
-      nodesDict: this.nodesDict,
+      nodesDict,
      type: this.type,
    };
  }
@@ -8,7 +8,7 @@ import {
  defaultKeywordExtractPrompt,
  defaultQueryKeywordExtractPrompt,
 } from "../../Prompt.js";
-import type { BaseRetriever } from "../../Retriever.js";
+import type { BaseRetriever, RetrieveParams } from "../../Retriever.js";
 import type { ServiceContext } from "../../ServiceContext.js";
 import { serviceContextFromDefaults } from "../../ServiceContext.js";
 import { RetrieverQueryEngine } from "../../engines/query/index.js";
@@ -79,7 +79,7 @@ abstract class BaseKeywordTableRetriever implements BaseRetriever {

  abstract getKeywords(query: string): Promise<string[]>;

-  async retrieve(query: string): Promise<NodeWithScore[]> {
+  async retrieve({ query }: RetrieveParams): Promise<NodeWithScore[]> {
    const keywords = await this.getKeywords(query);
    const chunkIndicesCount: { [key: string]: number } = {};
    const filteredKeywords = keywords.filter((keyword) =>
@@ -3,10 +3,9 @@ import { globalsHelper } from "../../GlobalsHelper.js";
 import type { BaseNode, Document, NodeWithScore } from "../../Node.js";
 import type { ChoiceSelectPrompt } from "../../Prompt.js";
 import { defaultChoiceSelectPrompt } from "../../Prompt.js";
-import type { BaseRetriever } from "../../Retriever.js";
+import type { BaseRetriever, RetrieveParams } from "../../Retriever.js";
 import type { ServiceContext } from "../../ServiceContext.js";
 import { serviceContextFromDefaults } from "../../ServiceContext.js";
-import type { Event } from "../../callbacks/CallbackManager.js";
 import { RetrieverQueryEngine } from "../../engines/query/index.js";
 import type { BaseNodePostprocessor } from "../../postprocessors/index.js";
 import type {
@@ -281,7 +280,10 @@ export class SummaryIndexRetriever implements BaseRetriever {
    this.index = index;
  }

-  async retrieve(query: string, parentEvent?: Event): Promise<NodeWithScore[]> {
+  async retrieve({
+    query,
+    parentEvent,
+  }: RetrieveParams): Promise<NodeWithScore[]> {
    const nodeIds = this.index.indexStruct.nodes;
    const nodes = await this.index.docStore.getNodes(nodeIds);
    const result = nodes.map((node) => ({
@@ -337,7 +339,10 @@ export class SummaryIndexLLMRetriever implements BaseRetriever {
    this.serviceContext = serviceContext || index.serviceContext;
  }

-  async retrieve(query: string, parentEvent?: Event): Promise<NodeWithScore[]> {
+  async retrieve({
+    query,
+    parentEvent,
+  }: RetrieveParams): Promise<NodeWithScore[]> {
    const nodeIds = this.index.indexStruct.nodes;
    const results: NodeWithScore[] = [];

@@ -11,7 +11,7 @@ import {
  ObjectType,
  splitNodesByType,
 } from "../../Node.js";
-import type { BaseRetriever } from "../../Retriever.js";
+import type { BaseRetriever, RetrieveParams } from "../../Retriever.js";
 import type { ServiceContext } from "../../ServiceContext.js";
 import { serviceContextFromDefaults } from "../../ServiceContext.js";
 import type { Event } from "../../callbacks/CallbackManager.js";
@@ -426,14 +426,17 @@ export class VectorIndexRetriever implements BaseRetriever {
    this.imageSimilarityTopK = imageSimilarityTopK ?? DEFAULT_SIMILARITY_TOP_K;
  }

-  async retrieve(
-    query: string,
-    parentEvent?: Event,
-    preFilters?: MetadataFilters,
-  ): Promise<NodeWithScore[]> {
-    let nodesWithScores = await this.textRetrieve(query, preFilters);
+  async retrieve({
+    query,
+    parentEvent,
+    preFilters,
+  }: RetrieveParams): Promise<NodeWithScore[]> {
+    let nodesWithScores = await this.textRetrieve(
+      query,
+      preFilters as MetadataFilters,
+    );
    nodesWithScores = nodesWithScores.concat(
-      await this.textToImageRetrieve(query, preFilters),
+      await this.textToImageRetrieve(query, preFilters as MetadataFilters),
    );
    this.sendEvent(query, nodesWithScores, parentEvent);
    return nodesWithScores;
@@ -1,7 +1,6 @@
 import type OpenAILLM from "openai";
 import type { ClientOptions as OpenAIClientOptions } from "openai";
 import type {
-  AnthropicStreamToken,
  CallbackManager,
  Event,
  EventType,
@@ -13,11 +12,7 @@ import type { ChatCompletionMessageParam } from "openai/resources/index.js";
 import type { LLMOptions } from "portkey-ai";
 import { Tokenizers, globalsHelper } from "../GlobalsHelper.js";
 import type { AnthropicSession } from "./anthropic.js";
-import {
-  ANTHROPIC_AI_PROMPT,
-  ANTHROPIC_HUMAN_PROMPT,
-  getAnthropicSession,
-} from "./anthropic.js";
+import { getAnthropicSession } from "./anthropic.js";
 import type { AzureOpenAIConfig } from "./azure.js";
 import {
  getAzureBaseUrl,
@@ -260,7 +255,7 @@ export class OpenAI extends BaseLLM {
      stream: false,
    });

-    const content = response.choices[0].message?.content ?? "";
+    const content = response.choices[0].message?.content ?? null;

    const kwargsOutput: Record<string, any> = {};

@@ -613,12 +608,30 @@ If a question does not make any sense, or is not factually coherent, explain why
  }
 }

-export const ALL_AVAILABLE_ANTHROPIC_MODELS = {
-  // both models have 100k context window, see https://docs.anthropic.com/claude/reference/selecting-a-model
-  "claude-2": { contextWindow: 200000 },
-  "claude-instant-1": { contextWindow: 100000 },
+export const ALL_AVAILABLE_ANTHROPIC_LEGACY_MODELS = {
+  "claude-2.1": {
+    contextWindow: 200000,
+  },
+  "claude-instant-1.2": {
+    contextWindow: 100000,
+  },
 };

+export const ALL_AVAILABLE_V3_MODELS = {
+  "claude-3-opus": { contextWindow: 200000 },
+  "claude-3-sonnet": { contextWindow: 200000 },
+};
+
+export const ALL_AVAILABLE_ANTHROPIC_MODELS = {
+  ...ALL_AVAILABLE_ANTHROPIC_LEGACY_MODELS,
+  ...ALL_AVAILABLE_V3_MODELS,
+};
+
+const AVAILABLE_ANTHROPIC_MODELS_WITHOUT_DATE: { [key: string]: string } = {
+  "claude-3-opus": "claude-3-opus-20240229",
+  "claude-3-sonnet": "claude-3-sonnet-20240229",
+} as { [key in keyof typeof ALL_AVAILABLE_ANTHROPIC_MODELS]: string };
+
 /**
 * Anthropic LLM implementation
 */
@@ -640,7 +653,7 @@ export class Anthropic extends BaseLLM {

  constructor(init?: Partial<Anthropic>) {
    super();
-    this.model = init?.model ?? "claude-2";
+    this.model = init?.model ?? "claude-3-opus";
    this.temperature = init?.temperature ?? 0.1;
    this.topP = init?.topP ?? 0.999; // Per Ben Mann
    this.maxTokens = init?.maxTokens ?? undefined;
@@ -674,21 +687,24 @@ export class Anthropic extends BaseLLM {
    };
  }

-  mapMessagesToPrompt(messages: ChatMessage[]) {
-    return (
-      messages.reduce((acc, message) => {
-        return (
-          acc +
-          `${
-            message.role === "system"
-              ? ""
-              : message.role === "assistant"
-                ? ANTHROPIC_AI_PROMPT + " "
-                : ANTHROPIC_HUMAN_PROMPT + " "
-          }${message.content.trim()}`
-        );
-      }, "") + ANTHROPIC_AI_PROMPT
-    );
+  getModelName = (model: string): string => {
+    if (Object.keys(AVAILABLE_ANTHROPIC_MODELS_WITHOUT_DATE).includes(model)) {
+      return AVAILABLE_ANTHROPIC_MODELS_WITHOUT_DATE[model];
+    }
+    return model;
+  };
+
+  formatMessages(messages: ChatMessage[]) {
+    return messages.map((message) => {
+      if (message.role !== "user" && message.role !== "assistant") {
+        throw new Error("Unsupported Anthropic role");
+      }
+
+      return {
+        content: message.content,
+        role: message.role,
+      };
+    });
  }

  chat(
@@ -698,49 +714,67 @@ export class Anthropic extends BaseLLM {
  async chat(
    params: LLMChatParamsNonStreaming | LLMChatParamsStreaming,
  ): Promise<ChatResponse | AsyncIterable<ChatResponseChunk>> {
-    const { messages, parentEvent, stream } = params;
+    let { messages } = params;
+
+    const { parentEvent, stream } = params;
+
+    let systemPrompt: string | null = null;
+
+    const systemMessages = messages.filter(
+      (message) => message.role === "system",
+    );
+
+    if (systemMessages.length > 0) {
+      systemPrompt = systemMessages
+        .map((message) => message.content)
+        .join("\n");
+      messages = messages.filter((message) => message.role !== "system");
+    }
+
    //Streaming
    if (stream) {
-      return this.streamChat(messages, parentEvent);
+      return this.streamChat(messages, parentEvent, systemPrompt);
    }

    //Non-streaming
-    const response = await this.session.anthropic.completions.create({
-      model: this.model,
-      prompt: this.mapMessagesToPrompt(messages),
-      max_tokens_to_sample: this.maxTokens ?? 100000,
+    const response = await this.session.anthropic.messages.create({
+      model: this.getModelName(this.model),
+      messages: this.formatMessages(messages),
+      max_tokens: this.maxTokens ?? 4096,
      temperature: this.temperature,
      top_p: this.topP,
+      ...(systemPrompt && { system: systemPrompt }),
    });

    return {
-      message: { content: response.completion.trimStart(), role: "assistant" },
-      //^ We're trimming the start because Anthropic often starts with a space in the response
-      // That space will be re-added when we generate the next prompt.
+      message: { content: response.content[0].text, role: "assistant" },
    };
  }

  protected async *streamChat(
    messages: ChatMessage[],
    parentEvent?: Event | undefined,
+    systemPrompt?: string | null,
  ): AsyncIterable<ChatResponseChunk> {
-    // AsyncIterable<AnthropicStreamToken>
-    const stream: AsyncIterable<AnthropicStreamToken> =
-      await this.session.anthropic.completions.create({
-        model: this.model,
-        prompt: this.mapMessagesToPrompt(messages),
-        max_tokens_to_sample: this.maxTokens ?? 100000,
-        temperature: this.temperature,
-        top_p: this.topP,
-        stream: true,
-      });
+    const stream = await this.session.anthropic.messages.create({
+      model: this.getModelName(this.model),
+      messages: this.formatMessages(messages),
+      max_tokens: this.maxTokens ?? 4096,
+      temperature: this.temperature,
+      top_p: this.topP,
+      stream: true,
+      ...(systemPrompt && { system: systemPrompt }),
+    });

    let idx_counter: number = 0;
    for await (const part of stream) {
-      //TODO: LLM Stream Callback, pending re-work.
+      const content =
+        part.type === "content_block_delta" ? part.delta.text : null;
+
+      if (typeof content !== "string") continue;

      idx_counter++;
-      yield { delta: part.completion };
+      yield { delta: content };
    }
    return;
  }
@@ -1,5 +1,6 @@
 import type { ClientOptions } from "@anthropic-ai/sdk";
 import Anthropic, { AI_PROMPT, HUMAN_PROMPT } from "@anthropic-ai/sdk";
+import { getEnv } from "@llamaindex/env";
 import _ from "lodash";

 export class AnthropicSession {
@@ -7,9 +8,7 @@ export class AnthropicSession {

  constructor(options: ClientOptions = {}) {
    if (!options.apiKey) {
-      if (typeof process !== undefined) {
-        options.apiKey = process.env.ANTHROPIC_API_KEY;
-      }
+      options.apiKey = getEnv("ANTHROPIC_API_KEY");
    }

    if (!options.apiKey) {
@@ -1,3 +1,5 @@
+import { getEnv } from "@llamaindex/env";
+
 export interface AzureOpenAIConfig {
  apiKey?: string;
  endpoint?: string;
@@ -67,24 +69,24 @@ export function getAzureConfigFromEnv(
  return {
    apiKey:
      init?.apiKey ??
-      process.env.AZURE_OPENAI_KEY ?? // From Azure docs
-      process.env.OPENAI_API_KEY ?? // Python compatible
-      process.env.AZURE_OPENAI_API_KEY, // LCJS compatible
+      getEnv("AZURE_OPENAI_KEY") ?? // From Azure docs
+      getEnv("OPENAI_API_KEY") ?? // Python compatible
+      getEnv("AZURE_OPENAI_API_KEY"), // LCJS compatible
    endpoint:
      init?.endpoint ??
-      process.env.AZURE_OPENAI_ENDPOINT ?? // From Azure docs
-      process.env.OPENAI_API_BASE ?? // Python compatible
-      process.env.AZURE_OPENAI_API_INSTANCE_NAME, // LCJS compatible
+      getEnv("AZURE_OPENAI_ENDPOINT") ?? // From Azure docs
+      getEnv("OPENAI_API_BASE") ?? // Python compatible
+      getEnv("AZURE_OPENAI_API_INSTANCE_NAME"), // LCJS compatible
    apiVersion:
      init?.apiVersion ??
-      process.env.AZURE_OPENAI_API_VERSION ?? // From Azure docs
-      process.env.OPENAI_API_VERSION ?? // Python compatible
-      process.env.AZURE_OPENAI_API_VERSION ?? // LCJS compatible
+      getEnv("AZURE_OPENAI_API_VERSION") ?? // From Azure docs
+      getEnv("OPENAI_API_VERSION") ?? // Python compatible
+      getEnv("AZURE_OPENAI_API_VERSION") ?? // LCJS compatible
      DEFAULT_API_VERSION,
    deploymentName:
      init?.deploymentName ??
-      process.env.AZURE_OPENAI_DEPLOYMENT ?? // From Azure docs
-      process.env.AZURE_OPENAI_API_DEPLOYMENT_NAME ?? // LCJS compatible
+      getEnv("AZURE_OPENAI_DEPLOYMENT") ?? // From Azure docs
+      getEnv("AZURE_OPENAI_API_DEPLOYMENT_NAME") ?? // LCJS compatible
      init?.model, // Fall back to model name, Python compatible
  };
 }
@@ -113,8 +115,8 @@ export function getAzureModel(openAIModel: string) {

 export function shouldUseAzure() {
  return (
-    process.env.AZURE_OPENAI_ENDPOINT ||
-    process.env.AZURE_OPENAI_API_INSTANCE_NAME ||
-    process.env.OPENAI_API_TYPE === "azure"
+    getEnv("AZURE_OPENAI_ENDPOINT") ||
+    getEnv("AZURE_OPENAI_API_INSTANCE_NAME") ||
+    getEnv("OPENAI_API_TYPE") === "azure"
  );
 }
@@ -1,9 +1,10 @@
+import { getEnv } from "@llamaindex/env";
 import { OpenAI } from "./LLM.js";

 export class FireworksLLM extends OpenAI {
  constructor(init?: Partial<OpenAI>) {
    const {
-      apiKey = process.env.FIREWORKS_API_KEY,
+      apiKey = getEnv("FIREWORKS_API_KEY"),
      additionalSessionOptions = {},
      model = "accounts/fireworks/models/mixtral-8x7b-instruct",
      ...rest
@@ -1,9 +1,10 @@
+import { getEnv } from "@llamaindex/env";
 import { OpenAI } from "./LLM.js";

 export class Groq extends OpenAI {
  constructor(init?: Partial<OpenAI>) {
    const {
-      apiKey = process.env.GROQ_API_KEY,
+      apiKey = getEnv("GROQ_API_KEY"),
      additionalSessionOptions = {},
      model = "mixtral-8x7b-32768",
      ...rest
@@ -1,3 +1,4 @@
+import { getEnv } from "@llamaindex/env";
 import type {
  CallbackManager,
  Event,
@@ -27,9 +28,7 @@ export class MistralAISession {
    if (init?.apiKey) {
      this.apiKey = init?.apiKey;
    } else {
-      if (typeof process !== undefined) {
-        this.apiKey = process.env.MISTRAL_API_KEY;
-      }
+      this.apiKey = getEnv("MISTRAL_API_KEY");
    }
    if (!this.apiKey) {
      throw new Error("Set Mistral API key in MISTRAL_API_KEY env variable"); // Overriding MistralAI package's error message
@@ -37,14 +37,18 @@ export class Ollama extends BaseEmbedding implements LLM {
  additionalChatOptions?: Record<string, unknown>;
  callbackManager?: CallbackManager;

+  protected modelMetadata: Partial<LLMMetadata>;
+
  constructor(
    init: Partial<Ollama> & {
      // model is required
      model: string;
+      modelMetadata?: Partial<LLMMetadata>;
    },
  ) {
    super();
    this.model = init.model;
+    this.modelMetadata = init.modelMetadata ?? {};
    Object.assign(this, init);
  }

@@ -56,6 +60,7 @@ export class Ollama extends BaseEmbedding implements LLM {
      maxTokens: undefined,
      contextWindow: this.contextWindow,
      tokenizer: undefined,
+      ...this.modelMetadata,
    };
  }

@@ -1,3 +1,4 @@
+import { getEnv } from "@llamaindex/env";
 import _ from "lodash";
 import type { ClientOptions } from "openai";
 import OpenAI from "openai";
@@ -13,9 +14,7 @@ export class OpenAISession {

  constructor(options: ClientOptions & { azure?: boolean } = {}) {
    if (!options.apiKey) {
-      if (typeof process !== undefined) {
-        options.apiKey = process.env.OPENAI_API_KEY;
-      }
+      options.apiKey = getEnv("OPENAI_API_KEY");
    }

    if (!options.apiKey) {
@@ -1,17 +1,8 @@
+import { getEnv } from "@llamaindex/env";
 import _ from "lodash";
 import type { LLMOptions } from "portkey-ai";
 import { Portkey } from "portkey-ai";

-export const readEnv = (
-  env: string,
-  default_val?: string,
-): string | undefined => {
-  if (typeof process !== "undefined") {
-    return process.env?.[env] ?? default_val;
-  }
-  return default_val;
-};
-
 interface PortkeyOptions {
  apiKey?: string;
  baseURL?: string;
@@ -24,11 +15,11 @@ export class PortkeySession {

  constructor(options: PortkeyOptions = {}) {
    if (!options.apiKey) {
-      options.apiKey = readEnv("PORTKEY_API_KEY");
+      options.apiKey = getEnv("PORTKEY_API_KEY");
    }

    if (!options.baseURL) {
-      options.baseURL = readEnv("PORTKEY_BASE_URL", "https://api.portkey.ai");
+      options.baseURL = getEnv("PORTKEY_BASE_URL") ?? "https://api.portkey.ai";
    }

    this.portkey = new Portkey({});
--- a/Show More
+++ b/Show More
Author	SHA1	Message	Date
Emanuel Ferreira	671c4432f8	chore: remove comment	2024-03-08 15:51:34 -03:00
Emanuel Ferreira	aefc3266c1	feat: experimental package + json query engine (#613 )	2024-03-07 14:34:55 -03:00
Huu Le (Lee)	fdf48dd459	feat: Add start in VSCode option and support python for dev container (#619 )	2024-03-07 17:19:08 +07:00
Alex Yang	66525346a2	build: use single swc config (#620 )	2024-03-06 23:41:42 -06:00
Alex Yang	c9b2ec4a2b	fix: release	2024-03-06 23:32:02 -06:00
Marcus Schiesser	bf583a7266	Use parameter object for retrieve function of Retriever (#616 )	2024-03-06 21:15:22 -08:00
Marcus Schiesser	de194d1c73	fix: running new-create-llama	2024-03-06 15:13:20 +07:00
Marcus Schiesser	ecdc289df1	RELEASING: Releasing 1 package(s) Releases: create-llama@0.0.27 [skip ci]	2024-03-06 15:11:47 +07:00
Marcus Schiesser	9e198ac40d	fix: build types for core locally (#615 )	2024-03-06 14:35:31 +07:00
Huu Le (Lee)	0a06998690	fix: hardcode "en" as default language for llama-parse and use llama cloud key from env (#614 )	2024-03-06 14:31:21 +07:00
Wojciech Grzebieniowski	484a7105a9	fix: restore missing exports (#610 )	2024-03-05 14:56:25 -06:00
Alex Yang	8d18ea167b	fix: publish.yml	2024-03-05 14:37:54 -06:00
Alex Yang	a2ca89bfe0	fix: config (#611 )	2024-03-05 14:20:58 -06:00
Alex Yang	edeea40898	ci: add publish.yml	2024-03-05 13:49:49 -06:00
Alex Yang	2a7080b094	build: fix version	2024-03-05 12:26:47 -06:00
Huu Le (Lee)	b354f2386b	feat: add embedding model option to create-llama (#608 )	2024-03-05 16:59:51 +07:00
Emanuel Ferreira	d766bd03d2	feat: OpenAI Agent Stream (#597 )	2024-03-05 15:46:44 +07:00
Huu Le (Lee)	6a69148356	fix: add --no-llama-parse and improve e2e test (#607 )	2024-03-05 14:57:38 +07:00
Marcus Schiesser	e1e1b0b522	RELEASING: Releasing 1 package(s) Releases: llamaindex@0.1.21 [skip ci]	2024-03-05 13:00:16 +07:00
Marcus Schiesser	d824876653	docs(changeset): Add support for Claude 3	2024-03-05 12:59:51 +07:00
Marcus Schiesser	2048698f77	docs: add interactive chat for anthropic	2024-03-05 11:15:50 +07:00
Emanuel Ferreira	9942979aa7	feat: Claude 3 (#604 )	2024-03-04 15:02:18 -08:00
Alex Yang	3c2655a1f9	fix: .tsbuildinfo	2024-03-04 16:05:45 -06:00
Marcus Schiesser	552a61a66f	Add quantized parameter to HuggingFaceEmbedding (#601 )	2024-03-04 12:10:40 +07:00
Alex Yang	d13143e322	RELEASING: Releasing 3 package(s) Releases: llamaindex@0.1.20 @llamaindex/env@0.0.5 docs@0.0.4 [skip ci]	2024-03-02 18:42:24 -06:00
Alex Yang	5116ad8d08	fix: compatibility issue with Deno (#598 )	2024-03-02 18:40:01 -06:00
Emanuel Ferreira	64683a55f3	fix: prefix messages always true (#596 )	2024-03-01 21:45:02 -03:00
Emanuel Ferreira	698cd9c631	fix: step wise agent + examples (#594 )	2024-03-01 21:28:02 -03:00
Alex Yang	c744a99102	chore: bump `@llamaindex/cloud` (#595 )	2024-03-01 17:22:50 -06:00
Huu Le (Lee)	2d2935085e	feat: Add use LlamaParse option to create-llama (#591 )	2024-03-01 16:54:06 +07:00
Marcus Schiesser	1b31e2c8cd	chore: update @llamaindex/cloud to 0.0.2	2024-03-01 15:59:10 +07:00
Thuc Pham	7257751993	fix: empty store bugs (#592 ) Co-authored-by: Marcus Schiesser <mail@marcusschiesser.de>	2024-03-01 15:04:10 +07:00
Marcus Schiesser	de6bfdb1b1	RELEASING: Releasing 1 package(s) Releases: llamaindex@0.1.19 [skip ci]	2024-02-29 15:35:13 +07:00
Marcus Schiesser	9e49f4411b	fix: copy README and license	2024-02-29 15:34:27 +07:00
Thuc Pham	026d068ddf	feat: enhance pinecone usage (#586 )	2024-02-29 15:34:08 +07:00
Marcus Schiesser	7055d6fc3c	docs: add OpenAIEmbedding to examples	2024-02-29 11:11:43 +07:00
Alex Yang	e9c2366bf1	fix: allow passing model metadata (#588 )	2024-02-29 10:41:06 +07:00
Alex Yang	6278152e49	fix: lazy import pg (#584 )	2024-02-27 19:16:54 -06:00
Emanuel Ferreira	76010c0cea	chore: remove duplicated example and minor example update (#582 )	2024-02-27 09:02:37 -03:00
Emanuel Ferreira	889b84cfb9	docs: remove query engine from correctness evaluator (#581 )	2024-02-27 08:15:41 -03:00