Guitaricet/relora

Official code for ReLoRA from the paper Stack More Layers Differently: High-Rank Training Through Low-Rank Updates

41
/ 100
Emerging

This tool helps machine learning engineers efficiently pre-train large language models. It takes preprocessed text data and a base model checkpoint, then trains the model using a technique that keeps memory usage low, producing a fine-tuned model ready for downstream tasks. It's designed for ML engineers working on custom language models.

474 stars. No commits in the last 6 months.

Use this if you need to pre-train a large language model with limited GPU memory but still want to achieve high-rank training performance.

Not ideal if you are looking for an out-of-the-box solution for inference or fine-tuning without needing to manage complex training configurations.

large-language-models model-pretraining deep-learning-optimization natural-language-processing
Stale 6m No Package No Dependents
Maintenance 0 / 25
Adoption 10 / 25
Maturity 16 / 25
Community 15 / 25

How are scores calculated?

Stars

474

Forks

42

Language

Jupyter Notebook

License

Apache-2.0

Last pushed

Apr 21, 2024

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/transformers/Guitaricet/relora"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.