forked from Archives/langchain
plutopulp
1 year ago
committed by
GitHub
parent
a35bbbfa9e
commit
6d6fd1b9e1
@ -0,0 +1,19 @@
|
||||
# PipelineAI
|
||||
|
||||
This page covers how to use the PipelineAI ecosystem within LangChain.
|
||||
It is broken into two parts: installation and setup, and then references to specific PipelineAI wrappers.
|
||||
|
||||
## Installation and Setup
|
||||
|
||||
- Install with `pip install pipeline-ai`
|
||||
- Get a Pipeline Cloud api key and set it as an environment variable (`PIPELINE_API_KEY`)
|
||||
|
||||
## Wrappers
|
||||
|
||||
### LLM
|
||||
|
||||
There exists a PipelineAI LLM wrapper, which you can access with
|
||||
|
||||
```python
|
||||
from langchain.llms import PipelineAI
|
||||
```
|
||||
@ -0,0 +1,171 @@
|
||||
{
|
||||
"cells": [
|
||||
{
|
||||
"attachments": {},
|
||||
"cell_type": "markdown",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"# PipelineAI\n",
|
||||
"\n",
|
||||
"PipelineAI allows you to run your ML models at scale in the cloud. It also provides API access to [several LLM models](https://pipeline.ai).\n",
|
||||
"\n",
|
||||
"This notebook goes over how to use Langchain with [PipelineAI](https://docs.pipeline.ai/docs)."
|
||||
]
|
||||
},
|
||||
{
|
||||
"attachments": {},
|
||||
"cell_type": "markdown",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## Install pipeline-ai\n",
|
||||
"The `pipeline-ai` library is required to use the `PipelineAI` API, AKA `Pipeline Cloud`. Install `pipeline-ai` using `pip install pipeline-ai`."
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"# Install the package\n",
|
||||
"!pip install pipeline-ai"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## Imports"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"import os\n",
|
||||
"from langchain.llms import PipelineAI\n",
|
||||
"from langchain import PromptTemplate, LLMChain"
|
||||
]
|
||||
},
|
||||
{
|
||||
"attachments": {},
|
||||
"cell_type": "markdown",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## Set the Environment API Key\n",
|
||||
"Make sure to get your API key from PipelineAI. Check out the [cloud quickstart guide](https://docs.pipeline.ai/docs/cloud-quickstart). You'll be given a 30 day free trial with 10 hours of serverless GPU compute to test different models."
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"os.environ[\"PIPELINE_API_KEY\"] = \"YOUR_API_KEY_HERE\""
|
||||
]
|
||||
},
|
||||
{
|
||||
"attachments": {},
|
||||
"cell_type": "markdown",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## Create the PipelineAI instance\n",
|
||||
"When instantiating PipelineAI, you need to specify the id or tag of the pipeline you want to use, e.g. `pipeline_key = \"public/gpt-j:base\"`. You then have the option of passing additional pipeline-specific keyword arguments:"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"llm = PipelineAI(pipeline_key=\"YOUR_PIPELINE_KEY\", pipeline_kwargs={...})"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## Create a Prompt Template\n",
|
||||
"We will create a prompt template for Question and Answer."
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"template = \"\"\"Question: {question}\n",
|
||||
"\n",
|
||||
"Answer: Let's think step by step.\"\"\"\n",
|
||||
"\n",
|
||||
"prompt = PromptTemplate(template=template, input_variables=[\"question\"])"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## Initiate the LLMChain"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"llm_chain = LLMChain(prompt=prompt, llm=llm)"
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "markdown",
|
||||
"metadata": {},
|
||||
"source": [
|
||||
"## Run the LLMChain\n",
|
||||
"Provide a question and run the LLMChain."
|
||||
]
|
||||
},
|
||||
{
|
||||
"cell_type": "code",
|
||||
"execution_count": null,
|
||||
"metadata": {},
|
||||
"outputs": [],
|
||||
"source": [
|
||||
"question = \"What NFL team won the Super Bowl in the year Justin Beiber was born?\"\n",
|
||||
"\n",
|
||||
"llm_chain.run(question)"
|
||||
]
|
||||
}
|
||||
],
|
||||
"metadata": {
|
||||
"kernelspec": {
|
||||
"display_name": "Python 3 (ipykernel)",
|
||||
"language": "python",
|
||||
"name": "python3"
|
||||
},
|
||||
"language_info": {
|
||||
"codemirror_mode": {
|
||||
"name": "ipython",
|
||||
"version": 3
|
||||
},
|
||||
"file_extension": ".py",
|
||||
"mimetype": "text/x-python",
|
||||
"name": "python",
|
||||
"nbconvert_exporter": "python",
|
||||
"pygments_lexer": "ipython3",
|
||||
"version": "3.10.6"
|
||||
},
|
||||
"vscode": {
|
||||
"interpreter": {
|
||||
"hash": "a0a0263b650d907a3bfe41c0f8d6a63a071b884df3cfdc1579f00cdc1aed6b03"
|
||||
}
|
||||
}
|
||||
},
|
||||
"nbformat": 4,
|
||||
"nbformat_minor": 4
|
||||
}
|
||||
@ -0,0 +1,107 @@
|
||||
"""Wrapper around Pipeline Cloud API."""
|
||||
import logging
|
||||
from typing import Any, Dict, List, Mapping, Optional
|
||||
|
||||
from pydantic import BaseModel, Extra, Field, root_validator
|
||||
|
||||
from langchain.llms.base import LLM
|
||||
from langchain.llms.utils import enforce_stop_tokens
|
||||
from langchain.utils import get_from_dict_or_env
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
|
||||
class PipelineAI(LLM, BaseModel):
|
||||
"""Wrapper around PipelineAI large language models.
|
||||
|
||||
To use, you should have the ``pipeline-ai`` python package installed,
|
||||
and the environment variable ``PIPELINE_API_KEY`` set with your API key.
|
||||
|
||||
Any parameters that are valid to be passed to the call can be passed
|
||||
in, even if not explicitly saved on this class.
|
||||
|
||||
Example:
|
||||
.. code-block:: python
|
||||
from langchain import PipelineAI
|
||||
pipeline = PipelineAI(pipeline_key="")
|
||||
"""
|
||||
|
||||
pipeline_key: str = ""
|
||||
"""The id or tag of the target pipeline"""
|
||||
|
||||
pipeline_kwargs: Dict[str, Any] = Field(default_factory=dict)
|
||||
"""Holds any pipeline parameters valid for `create` call not
|
||||
explicitly specified."""
|
||||
|
||||
pipeline_api_key: Optional[str] = None
|
||||
|
||||
class Config:
|
||||
"""Configuration for this pydantic config."""
|
||||
|
||||
extra = Extra.forbid
|
||||
|
||||
@root_validator(pre=True)
|
||||
def build_extra(cls, values: Dict[str, Any]) -> Dict[str, Any]:
|
||||
"""Build extra kwargs from additional params that were passed in."""
|
||||
all_required_field_names = {field.alias for field in cls.__fields__.values()}
|
||||
|
||||
extra = values.get("pipeline_kwargs", {})
|
||||
for field_name in list(values):
|
||||
if field_name not in all_required_field_names:
|
||||
if field_name in extra:
|
||||
raise ValueError(f"Found {field_name} supplied twice.")
|
||||
logger.warning(
|
||||
f"""{field_name} was transfered to pipeline_kwargs.
|
||||
Please confirm that {field_name} is what you intended."""
|
||||
)
|
||||
extra[field_name] = values.pop(field_name)
|
||||
values["pipeline_kwargs"] = extra
|
||||
return values
|
||||
|
||||
@root_validator()
|
||||
def validate_environment(cls, values: Dict) -> Dict:
|
||||
"""Validate that api key and python package exists in environment."""
|
||||
pipeline_api_key = get_from_dict_or_env(
|
||||
values, "pipeline_api_key", "PIPELINE_API_KEY"
|
||||
)
|
||||
values["pipeline_api_key"] = pipeline_api_key
|
||||
return values
|
||||
|
||||
@property
|
||||
def _identifying_params(self) -> Mapping[str, Any]:
|
||||
"""Get the identifying parameters."""
|
||||
return {
|
||||
**{"pipeline_key": self.pipeline_key},
|
||||
**{"pipeline_kwargs": self.pipeline_kwargs},
|
||||
}
|
||||
|
||||
@property
|
||||
def _llm_type(self) -> str:
|
||||
"""Return type of llm."""
|
||||
return "pipeline_ai"
|
||||
|
||||
def _call(self, prompt: str, stop: Optional[List[str]] = None) -> str:
|
||||
"""Call to Pipeline Cloud endpoint."""
|
||||
try:
|
||||
from pipeline import PipelineCloud
|
||||
except ImportError:
|
||||
raise ValueError(
|
||||
"Could not import pipeline-ai python package. "
|
||||
"Please install it with `pip install pipeline-ai`."
|
||||
)
|
||||
client = PipelineCloud(token=self.pipeline_api_key)
|
||||
params = self.pipeline_kwargs or {}
|
||||
|
||||
run = client.run_pipeline(self.pipeline_key, [prompt, params])
|
||||
try:
|
||||
text = run.result_preview[0][0]
|
||||
except AttributeError:
|
||||
raise AttributeError(
|
||||
f"A pipeline run should have a `result_preview` attribute."
|
||||
f"Run was: {run}"
|
||||
)
|
||||
if stop is not None:
|
||||
# I believe this is required since the stop tokens
|
||||
# are not enforced by the pipeline parameters
|
||||
text = enforce_stop_tokens(text, stop)
|
||||
return text
|
||||
@ -0,0 +1,10 @@
|
||||
"""Test Pipeline Cloud API wrapper."""
|
||||
|
||||
from langchain.llms.pipelineai import PipelineAI
|
||||
|
||||
|
||||
def test_pipelineai_call() -> None:
|
||||
"""Test valid call to Pipeline Cloud."""
|
||||
llm = PipelineAI()
|
||||
output = llm("Say foo:")
|
||||
assert isinstance(output, str)
|
||||
Loading…
Reference in New Issue