Callbacks Refactor [base] (#3256)

Co-authored-by: Nuno Campos <[email protected]> Co-authored-by: Davis Chase <[email protected]> Co-authored-by: Zander Chase <[email protected]> Co-authored-by: Harrison Chase <[email protected]>
langchain-ai · Apr 30, 2023 · d3ec00b · d3ec00b
1 parent 18ec22f
commit d3ec00b
Show file tree

Hide file tree

Showing 208 changed files with 6,297 additions and 3,256 deletions.
diff --git a/docs/ecosystem/aim_tracking.ipynb b/docs/ecosystem/aim_tracking.ipynb
@@ -61,7 +61,6 @@
     "from datetime import datetime\n",
     "\n",
     "from langchain.llms import OpenAI\n",
-    "from langchain.callbacks.base import CallbackManager\n",
     "from langchain.callbacks import AimCallbackHandler, StdOutCallbackHandler"
    ]
   },
@@ -109,8 +108,8 @@
     "    experiment_name=\"scenario 1: OpenAI LLM\",\n",
     ")\n",
     "\n",
-    "manager = CallbackManager([StdOutCallbackHandler(), aim_callback])\n",
-    "llm = OpenAI(temperature=0, callback_manager=manager, verbose=True)"
+    "callbacks = [StdOutCallbackHandler(), aim_callback]\n",
+    "llm = OpenAI(temperature=0, callbacks=callbacks)"
    ]
   },
   {
@@ -177,7 +176,7 @@
     "Title: {title}\n",
     "Playwright: This is a synopsis for the above play:\"\"\"\n",
     "prompt_template = PromptTemplate(input_variables=[\"title\"], template=template)\n",
-    "synopsis_chain = LLMChain(llm=llm, prompt=prompt_template, callback_manager=manager)\n",
+    "synopsis_chain = LLMChain(llm=llm, prompt=prompt_template, callbacks=callbacks)\n",
     "\n",
     "test_prompts = [\n",
     "    {\"title\": \"documentary about good video games that push the boundary of game design\"},\n",
@@ -249,13 +248,12 @@
    ],
    "source": [
     "# scenario 3 - Agent with Tools\n",
-    "tools = load_tools([\"serpapi\", \"llm-math\"], llm=llm, callback_manager=manager)\n",
+    "tools = load_tools([\"serpapi\", \"llm-math\"], llm=llm, callbacks=callbacks)\n",
     "agent = initialize_agent(\n",
     "    tools,\n",
     "    llm,\n",
     "    agent=AgentType.ZERO_SHOT_REACT_DESCRIPTION,\n",
-    "    callback_manager=manager,\n",
-    "    verbose=True,\n",
+    "    callbacks=callbacks,\n",
     ")\n",
     "agent.run(\n",
     "    \"Who is Leo DiCaprio's girlfriend? What is her current age raised to the 0.43 power?\"\n",

diff --git a/docs/ecosystem/clearml_tracking.ipynb b/docs/ecosystem/clearml_tracking.ipynb
@@ -79,7 +79,6 @@
    "source": [
     "from datetime import datetime\n",
     "from langchain.callbacks import ClearMLCallbackHandler, StdOutCallbackHandler\n",
-    "from langchain.callbacks.base import CallbackManager\n",
     "from langchain.llms import OpenAI\n",
     "\n",
     "# Setup and use the ClearML Callback\n",
@@ -93,9 +92,9 @@
     "    complexity_metrics=True,\n",
     "    stream_logs=True\n",
     ")\n",
-    "manager = CallbackManager([StdOutCallbackHandler(), clearml_callback])\n",
+    "callbacks = [StdOutCallbackHandler(), clearml_callback]\n",
     "# Get the OpenAI model ready to go\n",
-    "llm = OpenAI(temperature=0, callback_manager=manager, verbose=True)"
+    "llm = OpenAI(temperature=0, callbacks=callbacks)"
    ]
   },
   {
@@ -523,13 +522,12 @@
     "from langchain.agents import AgentType\n",
     "\n",
     "# SCENARIO 2 - Agent with Tools\n",
-    "tools = load_tools([\"serpapi\", \"llm-math\"], llm=llm, callback_manager=manager)\n",
+    "tools = load_tools([\"serpapi\", \"llm-math\"], llm=llm, callbacks=callbacks)\n",
     "agent = initialize_agent(\n",
     "    tools,\n",
     "    llm,\n",
     "    agent=AgentType.ZERO_SHOT_REACT_DESCRIPTION,\n",
-    "    callback_manager=manager,\n",
-    "    verbose=True,\n",
+    "    callbacks=callbacks,\n",
     ")\n",
     "agent.run(\n",
     "    \"Who is the wife of the person who sang summer of 69?\"\n",

diff --git a/docs/ecosystem/comet_tracking.ipynb b/docs/ecosystem/comet_tracking.ipynb
@@ -121,7 +121,6 @@
     "from datetime import datetime\n",
     "\n",
     "from langchain.callbacks import CometCallbackHandler, StdOutCallbackHandler\n",
-    "from langchain.callbacks.base import CallbackManager\n",
     "from langchain.llms import OpenAI\n",
     "\n",
     "comet_callback = CometCallbackHandler(\n",
@@ -131,8 +130,8 @@
     "    tags=[\"llm\"],\n",
     "    visualizations=[\"dep\"],\n",
     ")\n",
-    "manager = CallbackManager([StdOutCallbackHandler(), comet_callback])\n",
-    "llm = OpenAI(temperature=0.9, callback_manager=manager, verbose=True)\n",
+    "callbacks = [StdOutCallbackHandler(), comet_callback]\n",
+    "llm = OpenAI(temperature=0.9, callbacks=callbacks, verbose=True)\n",
     "\n",
     "llm_result = llm.generate([\"Tell me a joke\", \"Tell me a poem\", \"Tell me a fact\"] * 3)\n",
     "print(\"LLM result\", llm_result)\n",
@@ -153,7 +152,6 @@
    "outputs": [],
    "source": [
     "from langchain.callbacks import CometCallbackHandler, StdOutCallbackHandler\n",
-    "from langchain.callbacks.base import CallbackManager\n",
     "from langchain.chains import LLMChain\n",
     "from langchain.llms import OpenAI\n",
     "from langchain.prompts import PromptTemplate\n",
@@ -164,15 +162,14 @@
     "    stream_logs=True,\n",
     "    tags=[\"synopsis-chain\"],\n",
     ")\n",
-    "manager = CallbackManager([StdOutCallbackHandler(), comet_callback])\n",
-    "\n",
-    "llm = OpenAI(temperature=0.9, callback_manager=manager, verbose=True)\n",
+    "callbacks = [StdOutCallbackHandler(), comet_callback]\n",
+    "llm = OpenAI(temperature=0.9, callbacks=callbacks)\n",
     "\n",
     "template = \"\"\"You are a playwright. Given the title of play, it is your job to write a synopsis for that title.\n",
     "Title: {title}\n",
     "Playwright: This is a synopsis for the above play:\"\"\"\n",
     "prompt_template = PromptTemplate(input_variables=[\"title\"], template=template)\n",
-    "synopsis_chain = LLMChain(llm=llm, prompt=prompt_template, callback_manager=manager)\n",
+    "synopsis_chain = LLMChain(llm=llm, prompt=prompt_template, callbacks=callbacks)\n",
     "\n",
     "test_prompts = [{\"title\": \"Documentary about Bigfoot in Paris\"}]\n",
     "print(synopsis_chain.apply(test_prompts))\n",
@@ -194,7 +191,6 @@
    "source": [
     "from langchain.agents import initialize_agent, load_tools\n",
     "from langchain.callbacks import CometCallbackHandler, StdOutCallbackHandler\n",
-    "from langchain.callbacks.base import CallbackManager\n",
     "from langchain.llms import OpenAI\n",
     "\n",
     "comet_callback = CometCallbackHandler(\n",
@@ -203,15 +199,15 @@
     "    stream_logs=True,\n",
     "    tags=[\"agent\"],\n",
     ")\n",
-    "manager = CallbackManager([StdOutCallbackHandler(), comet_callback])\n",
-    "llm = OpenAI(temperature=0.9, callback_manager=manager, verbose=True)\n",
+    "callbacks = [StdOutCallbackHandler(), comet_callback]\n",
+    "llm = OpenAI(temperature=0.9, callbacks=callbacks)\n",
     "\n",
-    "tools = load_tools([\"serpapi\", \"llm-math\"], llm=llm, callback_manager=manager)\n",
+    "tools = load_tools([\"serpapi\", \"llm-math\"], llm=llm, callbacks=callbacks)\n",
     "agent = initialize_agent(\n",
     "    tools,\n",
     "    llm,\n",
     "    agent=\"zero-shot-react-description\",\n",
-    "    callback_manager=manager,\n",
+    "    callbacks=callbacks,\n",
     "    verbose=True,\n",
     ")\n",
     "agent.run(\n",
@@ -255,7 +251,6 @@
     "from rouge_score import rouge_scorer\n",
     "\n",
     "from langchain.callbacks import CometCallbackHandler, StdOutCallbackHandler\n",
-    "from langchain.callbacks.base import CallbackManager\n",
     "from langchain.chains import LLMChain\n",
     "from langchain.llms import OpenAI\n",
     "from langchain.prompts import PromptTemplate\n",
@@ -298,10 +293,10 @@
     "    tags=[\"custom_metrics\"],\n",
     "    custom_metrics=rouge_score.compute_metric,\n",
     ")\n",
-    "manager = CallbackManager([StdOutCallbackHandler(), comet_callback])\n",
-    "llm = OpenAI(temperature=0.9, callback_manager=manager, verbose=True)\n",
+    "callbacks = [StdOutCallbackHandler(), comet_callback]\n",
+    "llm = OpenAI(temperature=0.9)\n",
     "\n",
-    "synopsis_chain = LLMChain(llm=llm, prompt=prompt_template, callback_manager=manager)\n",
+    "synopsis_chain = LLMChain(llm=llm, prompt=prompt_template)\n",
     "\n",
     "test_prompts = [\n",
     "    {\n",
@@ -323,7 +318,7 @@
     "                 \"\"\"\n",
     "    }\n",
     "]\n",
-    "print(synopsis_chain.apply(test_prompts))\n",
+    "print(synopsis_chain.apply(test_prompts, callbacks=callbacks))\n",
     "comet_callback.flush_tracker(synopsis_chain, finish=True)"
    ]
   }

diff --git a/docs/ecosystem/gpt4all.md b/docs/ecosystem/gpt4all.md
@@ -3,6 +3,7 @@
 This page covers how to use the `GPT4All` wrapper within LangChain. The tutorial is divided into two parts: installation and setup, followed by usage with an example.
 
 ## Installation and Setup
+
 - Install the Python package with `pip install pyllamacpp`
 - Download a [GPT4All model](https://github.com/nomic-ai/pyllamacpp#supported-model) and place it in your desired directory
 
@@ -28,16 +29,16 @@ To stream the model's predictions, add in a CallbackManager.
 
 ```python
 from langchain.llms import GPT4All
-from langchain.callbacks.base import CallbackManager
 from langchain.callbacks.streaming_stdout import StreamingStdOutCallbackHandler
+
 # There are many CallbackHandlers supported, such as
 # from langchain.callbacks.streamlit import StreamlitCallbackHandler
 
-callback_manager = CallbackManager([StreamingStdOutCallbackHandler()])
-model = GPT4All(model="./models/gpt4all-model.bin", n_ctx=512, n_threads=8, callback_handler=callback_handler, verbose=True)
+callbacks = [StreamingStdOutCallbackHandler()]
+model = GPT4All(model="./models/gpt4all-model.bin", n_ctx=512, n_threads=8)
 
 # Generate text. Tokens are streamed through the callback manager.
-model("Once upon a time, ")
+model("Once upon a time, ", callbacks=callbacks)
 ```
 
 ## Model File

diff --git a/docs/ecosystem/wandb_tracking.ipynb b/docs/ecosystem/wandb_tracking.ipynb
@@ -50,7 +50,6 @@
    "source": [
     "from datetime import datetime\n",
     "from langchain.callbacks import WandbCallbackHandler, StdOutCallbackHandler\n",
-    "from langchain.callbacks.base import CallbackManager\n",
     "from langchain.llms import OpenAI"
    ]
   },
@@ -196,8 +195,8 @@
     "    name=\"llm\",\n",
     "    tags=[\"test\"],\n",
     ")\n",
-    "manager = CallbackManager([StdOutCallbackHandler(), wandb_callback])\n",
-    "llm = OpenAI(temperature=0, callback_manager=manager, verbose=True)"
+    "callbacks = [StdOutCallbackHandler(), wandb_callback]\n",
+    "llm = OpenAI(temperature=0, callbacks=callbacks)"
    ]
   },
   {
@@ -484,7 +483,7 @@
     "Title: {title}\n",
     "Playwright: This is a synopsis for the above play:\"\"\"\n",
     "prompt_template = PromptTemplate(input_variables=[\"title\"], template=template)\n",
-    "synopsis_chain = LLMChain(llm=llm, prompt=prompt_template, callback_manager=manager)\n",
+    "synopsis_chain = LLMChain(llm=llm, prompt=prompt_template, callbacks=callbacks)\n",
     "\n",
     "test_prompts = [\n",
     "    {\n",
@@ -577,16 +576,15 @@
    ],
    "source": [
     "# SCENARIO 3 - Agent with Tools\n",
-    "tools = load_tools([\"serpapi\", \"llm-math\"], llm=llm, callback_manager=manager)\n",
+    "tools = load_tools([\"serpapi\", \"llm-math\"], llm=llm)\n",
     "agent = initialize_agent(\n",
     "    tools,\n",
     "    llm,\n",
     "    agent=AgentType.ZERO_SHOT_REACT_DESCRIPTION,\n",
-    "    callback_manager=manager,\n",
-    "    verbose=True,\n",
     ")\n",
     "agent.run(\n",
-    "    \"Who is Leo DiCaprio's girlfriend? What is her current age raised to the 0.43 power?\"\n",
+    "    \"Who is Leo DiCaprio's girlfriend? What is her current age raised to the 0.43 power?\",\n",
+    "    callbacks=callbacks,\n",
     ")\n",
     "wandb_callback.flush_tracker(agent, reset=False, finish=True)"
    ]

diff --git a/docs/index.rst b/docs/index.rst
@@ -44,6 +44,8 @@ These modules are, in increasing order of complexity:
 
 - `Agents <./modules/agents.html>`_: Agents involve an LLM making decisions about which Actions to take, taking that Action, seeing an Observation, and repeating that until done. LangChain provides a standard interface for agents, a selection of agents to choose from, and examples of end to end agents.
 
+- `Callbacks <./modules/callbacks/getting_started.html>`_: It can be difficult to track all that occurs inside a chain or agent - callbacks help add a level of observability and introspection.
+
 
 .. toctree::
    :maxdepth: 1
@@ -57,6 +59,7 @@ These modules are, in increasing order of complexity:
    ./modules/memory.md
    ./modules/chains.md
    ./modules/agents.md
+   ./modules/callbacks/getting_started.ipynb
 
 Use Cases
 ----------