# tloen/alpaca-lora

Instruct-tune LLaMA on consumer hardware

Repository: https://github.com/tloen/alpaca-lora
Canonical: https://ross.abutalabs.com/products/alpaca-lora
Language: Jupyter Notebook
License: Apache-2.0
License Family: permissive
Last push: 2024-07-29T13:37:49+00:00

## Health v2 (maintenance only)
Score: 30/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 90
- inputs: {"age_days": 1269, "days_push": 765, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 18906, forks 2175 (observed 2026-08-28T04:11:27.196654+00:00)

## What it is
A repository of scripts for reproducing Stanford Alpaca-style instruction tuning of LLaMA models using low-rank adaptation (LoRA) on consumer GPUs. It includes training, inference, and weight-export code built on Hugging Face PEFT and bitsandbytes, plus pretrained LoRA weights.

## Use cases
- instruct-tune llama on a single consumer gpu
- fine-tune a large language model with lora
- run an alpaca-style chat model on a raspberry pi
- generate low-rank adapter weights for llama
- serve a fine-tuned llama model with a gradio interface
- merge lora weights into a base llama model

## When to choose
- you want to instruction-tune a 7b-65b llama model cheaply with LoRA
- you need a minimal reference implementation of PEFT-based fine-tuning
- you want pretrained Alpaca LoRA weights for inference on limited hardware

## When to avoid
- you need production-grade training infrastructure or multi-node training
- you want actively maintained support for newer model architectures
- you need a full fine-tuning pipeline rather than parameter-efficient adapters

## Facets
- artifact type: library
- maturity: maintenance
- function: llm-training, llm-inference, machine-learning, deep-learning
- domain: large-language-models, machine-learning, artificial-intelligence
- platform: python, cross-platform
- tags: lora, fine-tuning, llama, peft, bitsandbytes, consumer-hardware, gradio, jupyter-notebook, gpu, linux

## Member repositories
- tloen/alpaca-lora (main) score 30

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:11:27.196654+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T17:01:16.286874+00:00, confidence not recorded.
  - readme: https://github.com/tloen/alpaca-lora (fetched 2026-08-28T04:11:27.196654+00:00, sha b21fc0bcf43b)
- Data as of 2026-08-30T08:39:29.467469+00:00.