Added HuggingFaceTool to Explore HF Hub

[sk5268]

Purpose

I made a new tool openagi.actions.tools.huggingface.HuggingFaceTool that can search for models and datasets across the HuggingFace Hub.

Usage

Agent = Worker(
    role = "Model and Dataset searching agent",
    instructions = "Search for the prompted model or dataset",
    actions = [HuggingFaceTool]
)

Demo

import os
from openagi.agent import Admin
from openagi.memory import Memory
from openagi.worker import Worker
from openagi.llms.gemini import GeminiModel
from openagi.planner.task_decomposer import TaskPlanner
from openagi.actions.tools.hf import HuggingFaceTool

os.environ['GOOGLE_API_KEY'] = "YOUR_GEMINI_API_KEY"
os.environ['Gemini_MODEL'] = "gemini-1.5-flash"
os.environ['Gemini_TEMP'] = "0.5"

config = GeminiModel.load_from_env_config()
llm = GeminiModel(config=config)



# Implementing HuggingFace Agent

search_agent = Worker(
    role="Search Agent",
    instructions="""
    Search for the prompted model or dataset.
    """,
    actions=[HuggingFaceTool]
)

admin = Admin(
    llm=llm,
    planner=TaskPlanner(
        autonomous=False,
        human_intervene=False
    ),
    memory=Memory(),
    verbose=True
)

admin.assign_workers([
    search_agent
])

response = admin.run(
    query="gemma model",
    description="",
)

print(response)

Output

@lucifertrj

[tarun-aiplanet]

can you also extract other details as well. just model ID is very limited context given to the LLM

[sk5268]

The API returns the following details:

ModelInfo(id='prithivMLmods/Qwen2-VL-Math-Prase-2B-Instruct', author=None, sha=None,
created_at=datetime.datetime(2024, 12, 19, 4, 38, 46, tzinfo=datetime.timezone.utc),
last_modified=None, private=False, disabled=None, downloads=248, downloads_all_time=None,
gated=None, gguf=None, inference=None, likes=8, library_name='transformers',
tags=['transformers', 'safetensors', 'qwen2_vl', 'image-text-to-text', 'conversational', 'en',
'base_model:Qwen/Qwen2-VL-2B-Instruct', 'base_model:finetune:Qwen/Qwen2-VL-2B-Instruct',
'license:apache-2.0', 'text-generation-inference', 'endpoints_compatible', 'region:us'],
pipeline_tag='image-text-to-text', mask_token=None, card_data=None, widget_data=None, model_index=None,
config=None, transformers_info=None, trending_score=8, siblings=None, spaces=None,
safetensors=None, security_repo_status=None)

I think we can add:

• tags
• pipeline_tag
• config
• transfotmers_info

@tarun-aiplanet please let me know which one's to add and i'll make the changes.