Skip to content

lamalab-org/chempile-scripts

Repository files navigation

ChemPile scripts

This repository stores scripts for ChemPile:

  1. token_counter_cli.py -- script used to count tokens across the ChemPile subset.
  2. Submodule ddp_trainer:
    • Contains LoRA adapter merging scripts
    • ChemBench evaluation scripts
  3. Folder chempile-instruct contains the scripts used to generate the instruction dataset.
  4. review-app folder contains resources to reproduce the review app.

About

Scripts for ChemPile

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages