InflectionLM / CLAUDE.md
cafierom's picture
Upload 4 files
8788c39 verified
|
Raw
History Blame Contribute Delete
981 Bytes

A newer version of the Gradio SDK is available: 6.19.0

Upgrade

Project Overview

This repo contains code generating text using GEMMA 4 via the transformers library

Inflections

The goal of this work is to use GEMMA 4 to generate 5 beams in response to a prompt. Each beam is parsed and can be displayed along with it's total 'score.' Each beam can also be displayed with a score-per-token.

What needs work

  • create a second version of the inflections.py file with all class methods as separate functions. they should each return needed variables needed to call the next function,
  • modify the app.py so that the input elemnet takes the prompt and the output displays the first beam, with any token with score < 0.6 is highlighted in red.

Tech Stack

  • Language: Python
  • packages:
    • transformers
    • torch

Project Structure

llm_inflections/
  inflections.py           # inflection class
  app.py                   # gradio gui
  

Other information

  • assume any API keys can be obtained as environment variables.