unofficial-ffm-openai 0.2.1

An unofficial Formosa Foundation Model API client implementation compatible with LangChain and OpenAI

unofficial_ffm_openai_client

An unofficial Formosa Foundation Model client implementation based on OpenAI and LangChain

Introduction

This is an unofficial Python client implementation for the Formosa Foundation Model public endpoint, compatible with the OpenAI Python client and LangChain. Currently, it only implements the Conversation API and supports the public endpoint. Note that the synchronous API is not yet implemented.

Changelog

• 0.2.1 - [fix] Remove unnecessary import from chat_completion.py
• 0.2.0
  • Add a callback for counting token consumption in streaming and non-streaming mode.
  • Add json error handling for sync streaming mode.
  • Add token consumption info in the result of streaming and non-streaming mode.
• 0.1.3 - Support function calls.
• 0.1.2 - Support embeddings.

Usage

Install using pypi:

pip install unofficial-ffm-openai

You can use it similarly to the original OpenAIChat, with a few different parameters:

from ffm.langchain.language_models.ffm import FfmChatOpenAI

chat_ffm = FfmChatOpenAI(
    ffm_endpoint="https://api-ams.twcc.ai/api",
    max_tokens=1000,
    temperature=0.5,
    top_k=50,
    top_p=1.0,
    frequency_penalty=1.0,
    ffm_api_key="your key",
    ffm_deployment="ffm-mistral-7b-32k-instruct",  # or other model name
    streaming=True,
    callbacks=callbacks
)

from ffm.embeddings import FFMEmbeddings

embedding = FFMEmbeddings(
    base_url="",
    api_key="your key")

Callbacks

You cen use the ffm callbacks as using other callbacks such as openai callback, here is the example:

from ffm.langchain.callbacks import get_ffm_callback

with get_ffm_callback() as cb:
    ...do something using llm...
    
    total_tokens = cb.total_tokens
    prompt_tokens = cb.prompt_tokens
    completion_tokens = cb.completion_tokens
    successful_requests = cb.successful_requests

• Note: Cost calculation has not been done yet but is in progress.

Limitation

Currently, it has only been tested with the following dependencies:

langchain                         0.1.20
langchain-community               0.0.38
langchain-core                    0.1.52
langchain-openai                  0.1.7
langchain-text-splitters          0.0.2
langchainhub                      0.1.15

and the OpenAI client:

openai                            1.30.1

TODO

• Full implementation for the synchronous API.