Huggingface Tools#

Configuration File: huggingface_tools.json Tool Type: Local Tools Count: 3

This page contains all tools defined in the huggingface_tools.json configuration file.

Available Tools#

HuggingFace_get_model (Type: BaseRESTTool)#

Get detailed metadata for a specific model on HuggingFace Hub by its author and model name. Retur…

HuggingFace_get_model tool specification

Tool Information:

  • Name: HuggingFace_get_model

  • Type: BaseRESTTool

  • Description: Get detailed metadata for a specific model on HuggingFace Hub by its author and model name. Returns comprehensive information including model architecture, tags, download/like counts, creation date, license, widget data for interactive testing, and card metadata. Model IDs on HuggingFace follow the ‘author/model-name’ format; provide the author and model_name as separate parameters. Use HuggingFace_search_models to find model IDs (the ‘id’ field returns ‘author/model-name’).

Parameters:

  • author (string) (required) Model author or organization (e.g., ‘facebook’, ‘openai-community’, ‘google-bert’, ‘meta-llama’)

  • model_name (string) (required) Model name within the author’s namespace (e.g., ‘esm2_t33_650M_UR50D’, ‘gpt2’, ‘bert-base-uncased’)

Example Usage:

query = {
    "name": "HuggingFace_get_model",
    "arguments": {
        "author": "example_value",
        "model_name": "example_value"
    }
}
result = tu.run(query)

HuggingFace_search_datasets (Type: BaseRESTTool)#

Search HuggingFace Hub for machine learning datasets by keyword. Results use opaque cursor pagina…

HuggingFace_search_datasets tool specification

Tool Information:

  • Name: HuggingFace_search_datasets

  • Type: BaseRESTTool

  • Description: Search HuggingFace Hub for machine learning datasets by keyword. Results use opaque cursor pagination: begin without cursor and continue with metadata.next_cursor while metadata.has_more is true. The Hub does not expose an exact total for this endpoint. Returns dataset IDs, authors, downloads, likes, tags, and descriptions.

Parameters:

  • search (string) (required) Search query for dataset name or topic (e.g., ‘protein sequence’, ‘medical imaging’, ‘sentiment analysis’, ‘genomics’)

  • limit ([‘integer’, ‘null’]) (optional) Results per cursor page (default 20, max 1000)

  • cursor ([‘string’, ‘null’]) (optional) Opaque continuation cursor from metadata.next_cursor. Omit on the first page.

Example Usage:

query = {
    "name": "HuggingFace_search_datasets",
    "arguments": {
        "search": "example_value"
    }
}
result = tu.run(query)

HuggingFace_search_models (Type: BaseRESTTool)#

Search the HuggingFace Hub for machine learning models. Results use the Hub’s opaque cursor pagin…

HuggingFace_search_models tool specification

Tool Information:

  • Name: HuggingFace_search_models

  • Type: BaseRESTTool

  • Description: Search the HuggingFace Hub for machine learning models. Results use the Hub’s opaque cursor pagination: begin without cursor and continue with metadata.next_cursor while metadata.has_more is true. The Hub does not expose an exact total for this endpoint. Returns model IDs, downloads, likes, tags, pipeline task, and library information.

Parameters:

  • search (string) (required) Search query for model name, task, or topic (e.g., ‘protein language model’, ‘bert’, ‘image segmentation’, ‘drug discovery’)

  • limit ([‘integer’, ‘null’]) (optional) Results per cursor page (default 20, max 1000)

  • cursor ([‘string’, ‘null’]) (optional) Opaque continuation cursor from metadata.next_cursor. Omit on the first page.

  • pipeline_tag ([‘string’, ‘null’]) (optional) Filter by ML task type. Common values: ‘text-classification’, ‘token-classification’, ‘fill-mask’, ‘text-generation’, ‘image-classification’, ‘object-detection’, ‘feature-extraction’, ‘question-answering’, ‘summarization’, ‘translation’, ‘text-to-image’, ‘automatic-speech-recognition’

  • library ([‘string’, ‘null’]) (optional) Filter by ML framework/library. Common values: ‘transformers’, ‘diffusers’, ‘timm’, ‘sentence-transformers’, ‘spacy’, ‘flair’, ‘allennlp’, ‘adapter-transformers’

Example Usage:

query = {
    "name": "HuggingFace_search_models",
    "arguments": {
        "search": "example_value"
    }
}
result = tu.run(query)