Add ChatGLM3 llm usage

docs/docs/integrations/llms/chatglm.ipynb

@@ -11,7 +11,91 @@
     "\n",
     "[ChatGLM2-6B](https://github.com/THUDM/ChatGLM2-6B) is the second-generation version of the open-source bilingual (Chinese-English) chat model ChatGLM-6B. It retains the smooth conversation flow and low deployment threshold of the first-generation model, while introducing the new features like better performance, longer context and more efficient inference.\n",
     "\n",
-    "This example goes over how to use LangChain to interact with ChatGLM2-6B Inference for text completion.\n",
+    "[ChatGLM3](https://github.com/THUDM/ChatGLM3) is a new generation of pre-trained dialogue models jointly released by Zhipu AI and Tsinghua KEG. ChatGLM3-6B is the open-source model in the ChatGLM3 series"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## ChatGLM3\n",
+    "\n",
+    "This examples goes over how to use LangChain to interact with ChatGLM3-6B Inference for text completion."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from langchain.chains import LLMChain\n",
+    "from langchain.llms.chatglm3 import ChatGLM3\n",
+    "from langchain.prompts import PromptTemplate\n",
+    "from langchain.schema.messages import AIMessage"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "template = \"\"\"{question}\"\"\"\n",
+    "prompt = PromptTemplate(template=template, input_variables=[\"question\"])"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "endpoint_url = \"http://127.0.0.1:8000/v1/chat/completions\"\n",
+    "\n",
+    "messages = [\n",
+    "    AIMessage(content=\"我将从美国到中国来旅游，出行前希望了解中国的城市\"),\n",
+    "    AIMessage(content=\"欢迎问我任何问题。\"),\n",
+    "]\n",
+    "\n",
+    "llm = ChatGLM3(\n",
+    "    endpoint_url=endpoint_url,\n",
+    "    max_tokens=80000,\n",
+    "    prefix_messages=messages,\n",
+    "    top_p=0.9,\n",
+    ")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 4,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "'北京和上海是中国两个不同的城市,它们在很多方面都有所不同。\\n\\n北京是中国的首都,也是历史悠久的城市之一。它有着丰富的历史文化遗产,如故宫、颐和园等,这些景点吸引着众多游客前来观光。北京也是一个政治、文化和教育中心,有很多政府机构和学术机构总部设在北京。\\n\\n上海则是一个现代化的城市,它是中国的经济中心之一。上海拥有许多高楼大厦和国际化的金融机构,是中国最国际化的城市之一。上海也是一个美食和购物天堂,有许多著名的餐厅和购物中心。\\n\\n北京和上海的气候也不同。北京属于温带大陆性气候,冬季寒冷干燥,夏季炎热多风;而上海属于亚热带季风气候,四季分明,春秋宜人。\\n\\n北京和上海有很多不同之处,但都是中国非常重要的城市,每个城市都有自己独特的魅力和特色。'"
+      ]
+     },
+     "execution_count": 4,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "llm_chain = LLMChain(prompt=prompt, llm=llm)\n",
+    "question = \"北京和上海两座城市有什么不同？\"\n",
+    "\n",
+    "llm_chain.run(question)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "## ChatGLM and ChatGLM2\n",
+    "\n",
+    "The following example shows how to use LangChain to interact with the ChatGLM2-6B Inference to complete text.\n",
     "ChatGLM-6B and ChatGLM2-6B has the same api specs, so this example should work with both."
    ]
   },
@@ -106,7 +190,7 @@
  ],
  "metadata": {
   "kernelspec": {
-   "display_name": "langchain-dev",
+   "display_name": "Python 3 (ipykernel)",
    "language": "python",
    "name": "python3"
   },
@@ -120,9 +204,9 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.10.12"
+   "version": "3.11.5"
   }
  },
  "nbformat": 4,
- "nbformat_minor": 2
+ "nbformat_minor": 4
 }

libs/community/langchain_community/llms/chatglm3.py

@@ -0,0 +1,148 @@
+import json
+import logging
+from typing import Any, List, Optional, Union
+
+import httpx
+from langchain_core.callbacks import CallbackManagerForLLMRun
+from langchain_core.language_models.llms import LLM
+from langchain_core.messages import (
+    AIMessage,
+    BaseMessage,
+    FunctionMessage,
+    HumanMessage,
+    SystemMessage,
+)
+from langchain_core.pydantic_v1 import Field
+
+from langchain_community.llms.utils import enforce_stop_tokens
+
+logger = logging.getLogger(__name__)
+HEADERS = {"Content-Type": "application/json"}
+DEFAULT_TIMEOUT = 30
+
+
+def _convert_message_to_dict(message: BaseMessage) -> dict:
+    if isinstance(message, HumanMessage):
+        message_dict = {"role": "user", "content": message.content}
+    elif isinstance(message, AIMessage):
+        message_dict = {"role": "assistant", "content": message.content}
+    elif isinstance(message, SystemMessage):
+        message_dict = {"role": "system", "content": message.content}
+    elif isinstance(message, FunctionMessage):
+        message_dict = {"role": "function", "content": message.content}
+    else:
+        raise ValueError(f"Got unknown type {message}")
+    return message_dict
+
+
+class ChatGLM3(LLM):
+    """ChatGLM3 LLM service."""
+
+    model_name: str = Field(default="chatglm3-6b", alias="model")
+    endpoint_url: str = "http://127.0.0.1:8000/v1/chat/completions"
+    """Endpoint URL to use."""
+    model_kwargs: Optional[dict] = None
+    """Keyword arguments to pass to the model."""
+    max_tokens: int = 20000
+    """Max token allowed to pass to the model."""
+    temperature: float = 0.1
+    """LLM model temperature from 0 to 10."""
+    top_p: float = 0.7
+    """Top P for nucleus sampling from 0 to 1"""
+    prefix_messages: List[BaseMessage] = Field(default_factory=list)
+    """Series of messages for Chat input."""
+    streaming: bool = False
+    """Whether to stream the results or not."""
+    http_client: Union[Any, None] = None
+    timeout: int = DEFAULT_TIMEOUT
+
+    @property
+    def _llm_type(self) -> str:
+        return "chat_glm_3"
+
+    @property
+    def _invocation_params(self) -> dict:
+        """Get the parameters used to invoke the model."""
+        params = {
+            "model": self.model_name,
+            "temperature": self.temperature,
+            "max_tokens": self.max_tokens,
+            "top_p": self.top_p,
+            "stream": self.streaming,
+        }
+        return {**params, **(self.model_kwargs or {})}
+
+    @property
+    def client(self) -> Any:
+        return self.http_client or httpx.Client(timeout=self.timeout)
+
+    def _get_payload(self, prompt: str) -> dict:
+        params = self._invocation_params
+        messages = self.prefix_messages + [HumanMessage(content=prompt)]
+        params.update(
+            {
+                "messages": [_convert_message_to_dict(m) for m in messages],
+            }
+        )
+        return params
+
+    def _call(
+        self,
+        prompt: str,
+        stop: Optional[List[str]] = None,
+        run_manager: Optional[CallbackManagerForLLMRun] = None,
+        **kwargs: Any,
+    ) -> str:
+        """Call out to a ChatGLM3 LLM inference endpoint.
+
+        Args:
+            prompt: The prompt to pass into the model.
+            stop: Optional list of stop words to use when generating.
+
+        Returns:
+            The string generated by the model.
+
+        Example:
+            .. code-block:: python
+
+                response = chatglm_llm("Who are you?")
+        """
+        payload = self._get_payload(prompt)
+        logger.debug(f"ChatGLM3 payload: {payload}")
+
+        try:
+            response = self.client.post(
+                self.endpoint_url, headers=HEADERS, json=payload
+            )
+        except httpx.NetworkError as e:
+            raise ValueError(f"Error raised by inference endpoint: {e}")
+
+        logger.debug(f"ChatGLM3 response: {response}")
+
+        if response.status_code != 200:
+            raise ValueError(f"Failed with response: {response}")
+
+        try:
+            parsed_response = response.json()
+
+            if isinstance(parsed_response, dict):
+                content_keys = "choices"
+                if content_keys in parsed_response:
+                    choices = parsed_response[content_keys]
+                    if len(choices):
+                        text = choices[0]["message"]["content"]
+                else:
+                    raise ValueError(f"No content in response : {parsed_response}")
+            else:
+                raise ValueError(f"Unexpected response type: {parsed_response}")
+
+        except json.JSONDecodeError as e:
+            raise ValueError(
+                f"Error raised during decoding response from inference endpoint: {e}."
+                f"\nResponse: {response.text}"
+            )
+
+        if stop is not None:
+            text = enforce_stop_tokens(text, stop)
+
+        return text

libs/community/pyproject.toml

@@ -83,6 +83,7 @@ msal = {version = "^1.25.0", optional = true}
 databricks-vectorsearch = {version = "^0.21", optional = true}
 dgml-utils = {version = "^0.3.0", optional = true}
 datasets = {version = "^2.15.0", optional = true}
+httpx = "^0.24.1"
 
 [tool.poetry.group.test]
 optional = true

libs/community/tests/unit_tests/test_dependencies.py

@@ -41,6 +41,7 @@ def test_required_dependencies(poetry_conf: Mapping[str, Any]) -> None:
             "SQLAlchemy",
             "aiohttp",
             "dataclasses-json",
+            "httpx",
             "langchain-core",
             "langsmith",
             "numpy",

libs/langchain/langchain/llms/chatglm3.py

@@ -0,0 +1,3 @@
+from langchain_community.llms.chatglm3 import ChatGLM3
+
+__all__ = ["ChatGLM3"]

libs/langchain/pyproject.toml

@@ -22,6 +22,7 @@ numpy = "^1"
 aiohttp = "^3.8.3"
 tenacity = "^8.1.0"
 jsonpatch = "^1.33"
+httpx = "^0.24.1"
 azure-core = {version = "^1.26.4", optional=true}
 tqdm = {version = ">=4.48.0", optional = true}
 openapi-pydantic = {version = "^0.3.2", optional = true}

libs/langchain/tests/unit_tests/test_dependencies.py

@@ -42,6 +42,7 @@ def test_required_dependencies(poetry_conf: Mapping[str, Any]) -> None:
             "aiohttp",
             "async-timeout",
             "dataclasses-json",
+            "httpx",
             "jsonpatch",
             "langchain-core",
             "langsmith",