annotateai

Here is a more detailed and structured version of the documentation:

Introduction

The annotate command-line tool is used to annotate papers with concepts. It supports various LLMs, including those from Hugging Face Hub.

Usage

To use the tool, run the following command:

annotate <paper_url> [options]

Replace <paper_url> with the URL of the paper you want to annotate.

Options

• --keywords : Specify a list of keywords to search for in the paper.
• --progress : Disable the progress bar. (default: True)
• --LLM : Set the LLM to use. Can be set via ENV parameters or the Hugging Face Hub.

Supported LLMs

The following LLMs are currently supported:

• gpt-4o
• claude-3-5-sonnet-20240620
• ollama/llama3.1
• bartowski/Llama-3.1_OpenScholar-8B-GGUF/Llama-3.1_OpenScholar-8B-Q4_K_M.gguf

Docker Web Application

A web application version of the tool is available on Docker Hub.

To run the web application, use the following command:

docker run -d --gpus=all -it -p 8501:8501 neuml/annotateai

The LLM can also be set via ENV parameters. For example:

docker run -d --gpus=all -it -p 8501:8501 -e LLM=bartowski/Llama-3.2-1B-Instruct-GGUF/Llama-3.2-1B-Instruct-Q4_K_M.gguf neuml/annotateai

The code for the web application can be found in the app folder.

API Documentation

The tool provides the following API endpoints:

• POST /annotate: Annotate a paper with concepts.
  • Request Body: Paper URL, keywords (optional), progress (optional).
  • Response: JSON object containing the annotation results.

Example request:

curl -X POST \
  http://localhost:8501/annotate \
  -H 'Content-Type: application/json' \
  -d '{"paper_url": "https://arxiv.org/pdf/2005.11401", "keywords": ["hallucinations", "llm"], "progress": false}'

LLM APIs

The tool also provides the following LLM API endpoints:

• GET /llms: Get a list of supported LLMs.
  • Response: JSON array containing the LLM names.

Example request:

curl -X GET \
  http://localhost:8501/llms

• POST /llms: Set the LLM to use.
  • Request Body: LLM name or ENV variable (e.g. LLM=bartowski/Llama-3.2-1B-Instruct-GGUF/Llama-3.2-1B-Instruct-Q4_K_M.gguf).
  • Response: JSON object containing the LLM settings.

Example request:

curl -X POST \
  http://localhost:8501/llms \
  -H 'Content-Type: application/json' \
  -d '{"llm_name": "bartowski/Llama-3.2-1B-Instruct-GGUF/Llama-3.2-1B-Instruct-Q4_K_M.gguf"}'

Note that this is a more detailed and structured version of the documentation, including API endpoints and LLM APIs.