llm_lib

Gem to invoke API calls to Huggingface and OpenAI LLMs.

Support both Huggingface and OpenAI models.

Install

gem install llm_lib

Usage examples

Replace 'API_KEY' with your actual OpenAI/Huggingface API key

OpenAI API call

apikey = ENV['OPENAI_API_KEY']
client = LlmLib::OpenAIClient.new(apikey)

prompt = "Once upon a time"
max_tokens = 100
response = client.chat_gpt_call(prompt, 
                        max_tokens)
puts response

HuggingFace API call

apikey = ENV['HUGGINGFACE_API_KEY']
client = LlmLib::HuggingfaceApiClient.new(apikey)

query = "Once upon a time"
response = client.hugging_bloom_call(query)

puts response

Supported models

Currently support following models, more models will be added in the future.

OpenAI

• gpt-3.5-turbo
• gpt-4

Huggingface

• bigscience/bloom
• tiiuae/falcon-40b-instruct
• meta-llama/Llama-2-70b-chat-hf
• databricks/dolly-v2-12b
• google/flan-t5-xl

Custom model

Use following example to invoke unavailable models from huggingface.

class HuggingfaceApiClient

    def initialize(api_key)
        @api_key = api_key
    end

    def hugging_dolly2_call(query, model = "databricks/dolly-v2-12b")
        response = HuggingFace.send(@api_key,
                                model,
                                query 
                                )
        response
    end
end

require 'llm_lib/huggingface'

For more visit the documentation.

Disclaimer

This is only a API wrapper for OpenAI and Huggingface models. Usage license, pricing, limitations, etc; are relevant to each model and provider, thus their guidelines and documentation must be read and followed before using.