Merge branch 'main' into opensearch-binary-byte-support

run-llama · Jan 23, 2025 · 5804236 · 5804236
2 parents e56a7ca + 2d35a0f
commit 5804236
Show file tree

Hide file tree

Showing 29 changed files with 799 additions and 148 deletions.
diff --git a/.github/workflows/publish_release.yml b/.github/workflows/publish_release.yml
@@ -38,7 +38,7 @@ jobs:
         shell: bash
         run: rm -rf llama-index-core/llama_index/core/_static/nltk_cache/corpora/stopwords.zip llama-index-core/llama_index/core/_static/nltk_cache/tokenizers/punkt.zip
       - name: Build and publish to pypi
-        uses: JRubics/poetry-publish@v2.0
+        uses: JRubics/poetry-publish@v2.1
         with:
           python_version: ${{ env.PYTHON_VERSION }}
           pypi_token: ${{ secrets.LLAMA_INDEX_PYPI_TOKEN }}

diff --git a/docs/docs/api_reference/tools/linkup_research.md b/docs/docs/api_reference/tools/linkup_research.md
@@ -0,0 +1,4 @@
+::: llama_index.tools.tavily_research
+    options:
+      members:
+        - LinkupToolSpec
diff --git a/docs/docs/examples/agent/react_agent.ipynb b/docs/docs/examples/agent/react_agent.ipynb
@@ -53,22 +53,7 @@
    "execution_count": null,
    "id": "e8ac1778-0585-43c9-9dad-014d13d7460d",
    "metadata": {},
-   "outputs": [
-    {
-     "name": "stderr",
-     "output_type": "stream",
-     "text": [
-      "[nltk_data] Downloading package stopwords to /Users/jerryliu/Programmi\n",
-      "[nltk_data]     ng/gpt_index/.venv/lib/python3.10/site-\n",
-      "[nltk_data]     packages/llama_index/legacy/_static/nltk_cache...\n",
-      "[nltk_data]   Unzipping corpora/stopwords.zip.\n",
-      "[nltk_data] Downloading package punkt to /Users/jerryliu/Programming/g\n",
-      "[nltk_data]     pt_index/.venv/lib/python3.10/site-\n",
-      "[nltk_data]     packages/llama_index/legacy/_static/nltk_cache...\n",
-      "[nltk_data]   Unzipping tokenizers/punkt.zip.\n"
-     ]
-    }
-   ],
+   "outputs": [],
    "source": [
     "from llama_index.core.agent import ReActAgent\n",
     "from llama_index.llms.openai import OpenAI\n",
@@ -468,13 +453,44 @@
     "response = agent.chat(\"What is 5+3+2\")\n",
     "print(response)"
    ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "76190511-692c-4642-9b86-adac88c98550",
+   "metadata": {},
+   "source": [
+    "### Customizing the Message Role of Observation\n",
+    "\n",
+    "If the LLM you use supports function/tool calling, you may set the message role of observations to `MessageRole.TOOL`.  \n",
+    "Doing this will prevent the tool outputs from being misinterpreted as new user messages for some models."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "e6d5e8c1-c40e-4a96-8d2e-84127f066265",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from llama_index.core.agent import ReActChatFormatter\n",
+    "from llama_index.core.llms import MessageRole\n",
+    "\n",
+    "agent = ReActAgent.from_tools(\n",
+    "    [multiply_tool, add_tool],\n",
+    "    llm=llm,\n",
+    "    react_chat_formatter=ReActChatFormatter.from_defaults(\n",
+    "        observation_role=MessageRole.TOOL\n",
+    "    ),\n",
+    "    verbose=True,\n",
+    ")"
+   ]
   }
  ],
  "metadata": {
   "kernelspec": {
-   "display_name": "llama_index_v2",
+   "display_name": "LlamaIndex Development",
    "language": "python",
-   "name": "llama_index_v2"
+   "name": "llama-index-dev"
   },
   "language_info": {
    "codemirror_mode": {

diff --git a/llama-index-cli/llama_index/cli/upgrade/mappings.json b/llama-index-cli/llama_index/cli/upgrade/mappings.json
@@ -534,6 +534,7 @@
   "AzureSpeechToolSpec": "llama_index.tools.azure_speech",
   "WolframAlphaToolSpec": "llama_index.tools.wolfram_alpha",
   "SlackToolSpec": "llama_index.tools.slack",
+  "LinkupToolSpec": "llama_index.tools.linkup_research",
   "TavilyToolSpec": "llama_index.tools.tavily_research",
   "ArxivToolSpec": "llama_index.tools.arxiv",
   "VectorDB": "llama_index.tools.vector_db",

diff --git a/llama-index-core/llama_index/core/agent/react/formatter.py b/llama-index-core/llama_index/core/agent/react/formatter.py
@@ -13,7 +13,7 @@
     ObservationReasoningStep,
 )
 from llama_index.core.base.llms.types import ChatMessage, MessageRole
-from llama_index.core.bridge.pydantic import BaseModel, ConfigDict
+from llama_index.core.bridge.pydantic import BaseModel, ConfigDict, Field
 from llama_index.core.tools import BaseTool
 
 logger = logging.getLogger(__name__)
@@ -53,6 +53,14 @@ class ReActChatFormatter(BaseAgentChatFormatter):
 
     system_header: str = REACT_CHAT_SYSTEM_HEADER  # default
     context: str = ""  # not needed w/ default
+    observation_role: MessageRole = Field(
+        default=MessageRole.USER,
+        description=(
+            "Message role of tool outputs. If the LLM you use supports function/tool "
+            "calling, you may set it to `MessageRole.TOOL` to avoid the tool outputs "
+            "being misinterpreted as new user messages."
+        ),
+    )
 
     def format(
         self,
@@ -73,13 +81,13 @@ def format(
         fmt_sys_header = self.system_header.format(**format_args)
 
         # format reasoning history as alternating user and assistant messages
-        # where the assistant messages are thoughts and actions and the user
+        # where the assistant messages are thoughts and actions and the tool
         # messages are observations
         reasoning_history = []
         for reasoning_step in current_reasoning:
             if isinstance(reasoning_step, ObservationReasoningStep):
                 message = ChatMessage(
-                    role=MessageRole.USER,
+                    role=self.observation_role,
                     content=reasoning_step.get_content(),
                 )
             else:
@@ -100,6 +108,7 @@ def from_defaults(
         cls,
         system_header: Optional[str] = None,
         context: Optional[str] = None,
+        observation_role: MessageRole = MessageRole.USER,
     ) -> "ReActChatFormatter":
         """Create ReActChatFormatter from defaults."""
         if not system_header:
@@ -112,6 +121,7 @@ def from_defaults(
         return ReActChatFormatter(
             system_header=system_header,
             context=context or "",
+            observation_role=observation_role,
         )
 
     @classmethod

diff --git a/llama-index-core/llama_index/core/agent/workflow/multi_agent_workflow.py b/llama-index-core/llama_index/core/agent/workflow/multi_agent_workflow.py
@@ -472,7 +472,7 @@ def from_tools_or_functions(
 
         tools = [
             FunctionTool.from_defaults(fn=tool)
-            if not isinstance(tool, FunctionTool)
+            if not isinstance(tool, BaseTool)
             else tool
             for tool in tools_or_functions
         ]

diff --git a/llama-index-core/llama_index/core/command_line/mappings.json b/llama-index-core/llama_index/core/command_line/mappings.json
@@ -533,6 +533,7 @@
   "AzureSpeechToolSpec": "llama_index.tools.azure_speech",
   "WolframAlphaToolSpec": "llama_index.tools.wolfram_alpha",
   "SlackToolSpec": "llama_index.tools.slack",
+  "LinkupToolSpec": "llama_index.tools.linkup_research",
   "TavilyToolSpec": "llama_index.tools.tavily_research",
   "ArxivToolSpec": "llama_index.tools.arxiv",
   "VectorDB": "llama_index.tools.vector_db",

diff --git a/llama-index-core/llama_index/core/prompts/base.py b/llama-index-core/llama_index/core/prompts/base.py
@@ -47,6 +47,7 @@
 from llama_index.core.base.llms.generic_utils import (
     prompt_to_messages,
 )
+from llama_index.core.base.llms.types import ContentBlock, TextBlock
 from llama_index.core.prompts.prompt_type import PromptType
 from llama_index.core.prompts.utils import get_template_vars, format_string
 from llama_index.core.types import BaseOutputParser
@@ -305,20 +306,30 @@ def format_messages(
 
         messages: List[ChatMessage] = []
         for message_template in self.message_templates:
-            message_content = message_template.content or ""
-
-            template_vars = get_template_vars(message_content)
-            relevant_kwargs = {
-                k: v for k, v in mapped_all_kwargs.items() if k in template_vars
-            }
-            content_template = message_template.content or ""
-
-            # if there's mappings specified, make sure those are used
-            content = format_string(content_template, **relevant_kwargs)
-
-            message: ChatMessage = message_template.model_copy()
-            message.content = content
-            messages.append(message)
+            # Handle messages with multiple blocks
+            if message_template.blocks:
+                formatted_blocks: List[ContentBlock] = []
+                for block in message_template.blocks:
+                    if isinstance(block, TextBlock):
+                        template_vars = get_template_vars(block.text)
+                        relevant_kwargs = {
+                            k: v
+                            for k, v in mapped_all_kwargs.items()
+                            if k in template_vars
+                        }
+                        formatted_text = format_string(block.text, **relevant_kwargs)
+                        formatted_blocks.append(TextBlock(text=formatted_text))
+                    else:
+                        # For non-text blocks (like images), keep them as is
+                        # TODO: can images be formatted as variables?
+                        formatted_blocks.append(block)
+
+                message = message_template.model_copy()
+                message.blocks = formatted_blocks
+                messages.append(message)
+            else:
+                # Handle empty messages (if any)
+                messages.append(message_template.model_copy())
 
         if self.output_parser is not None:
             messages = self.output_parser.format_messages(messages)

diff --git a/llama-index-integrations/llms/llama-index-llms-mistralai/llama_index/llms/mistralai/utils.py b/llama-index-integrations/llms/llama-index-llms-mistralai/llama_index/llms/mistralai/utils.py
@@ -22,6 +22,9 @@
     "open-mixtral-8x22b",
     "ministral-8b-latest",
     "ministral-3b-latest",
+    "mistral-small-latest",
+    "codestral-latest",
+    "open-mistral-nemo-latest",
 )
 
 MISTRALAI_CODE_MODELS = "codestral-latest"

diff --git a/llama-index-integrations/llms/llama-index-llms-mistralai/pyproject.toml b/llama-index-integrations/llms/llama-index-llms-mistralai/pyproject.toml
@@ -27,7 +27,7 @@ exclude = ["**/BUILD"]
 license = "MIT"
 name = "llama-index-llms-mistralai"
 readme = "README.md"
-version = "0.3.1"
+version = "0.3.2"
 
 [tool.poetry.dependencies]
 python = ">=3.9,<4.0"

diff --git a/llama-index-integrations/tools/llama-index-tools-linkup-research/.gitignore b/llama-index-integrations/tools/llama-index-tools-linkup-research/.gitignore
@@ -0,0 +1,153 @@
+llama_index/_static
+.DS_Store
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[cod]
+*$py.class
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+bin/
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+etc/
+include/
+lib/
+lib64/
+parts/
+sdist/
+share/
+var/
+wheels/
+pip-wheel-metadata/
+share/python-wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# PyInstaller
+#  Usually these files are written by a python script from a template
+#  before PyInstaller builds the exe, so as to inject date/other infos into it.
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.nox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+*.py,cover
+.hypothesis/
+.pytest_cache/
+.ruff_cache
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+db.sqlite3
+db.sqlite3-journal
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# Sphinx documentation
+docs/_build/
+
+# PyBuilder
+target/
+
+# Jupyter Notebook
+.ipynb_checkpoints
+notebooks/
+
+# IPython
+profile_default/
+ipython_config.py
+
+# pyenv
+.python-version
+
+# pipenv
+#   According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
+#   However, in case of collaboration, if having platform-specific dependencies or dependencies
+#   having no cross-platform support, pipenv may install dependencies that don't work, or not
+#   install all needed dependencies.
+#Pipfile.lock
+
+# PEP 582; used by e.g. github.com/David-OConnor/pyflow
+__pypackages__/
+
+# Celery stuff
+celerybeat-schedule
+celerybeat.pid
+
+# SageMath parsed files
+*.sage.py
+
+# Environments
+.env
+.venv
+env/
+venv/
+ENV/
+env.bak/
+venv.bak/
+pyvenv.cfg
+
+# Spyder project settings
+.spyderproject
+.spyproject
+
+# Rope project settings
+.ropeproject
+
+# mkdocs documentation
+/site
+
+# mypy
+.mypy_cache/
+.dmypy.json
+dmypy.json
+
+# Pyre type checker
+.pyre/
+
+# Jetbrains
+.idea
+modules/
+*.swp
+
+# VsCode
+.vscode
+
+# pipenv
+Pipfile
+Pipfile.lock
+
+# pyright
+pyrightconfig.json
diff --git a/llama-index-integrations/tools/llama-index-tools-linkup-research/BUILD b/llama-index-integrations/tools/llama-index-tools-linkup-research/BUILD
@@ -0,0 +1,3 @@
+poetry_requirements(
+    name="poetry",
+)