Home Esplora Aiuto
Accedi
dongzenan
/
alpaca-lora
1
0
Forka 0
File Problemi 0 Pull Requests 0 Wiki

Nessuna descrizione

Albero (Tree): 63121244c8
Rami (Branch) Tag
alpaca-lora
Eric Wang 63121244c8 Licenses and whatnot 3 anni fa
.gitignore 26f64780ad initial commit 3 anni fa
DATA_LICENSE 63121244c8 Licenses and whatnot 3 anni fa
LICENSE 63121244c8 Licenses and whatnot 3 anni fa
README.md 63121244c8 Licenses and whatnot 3 anni fa
alpaca_data.json 26f64780ad initial commit 3 anni fa
conversion.py 26f64780ad initial commit 3 anni fa
finetune.py 63121244c8 Licenses and whatnot 3 anni fa
generate.py 63121244c8 Licenses and whatnot 3 anni fa
iteration.ipynb 26f64780ad initial commit 3 anni fa
lengths.ipynb 26f64780ad initial commit 3 anni fa

README.md

alpaca-lora

This repository contains code for reproducing the Stanford Alpaca results. Users will need to have LLaMA weights on hand and be ready to fork transformers.

  1. Install dependencies

    pip install -q bitsandbytes datasets accelerate loralib

pip install -q git+https://github.com/zphang/transformers@llama_push
pip install -q git+https://github.com/huggingface/peft.git\

  2. Convert weights

    python conversion.py --input_dir [LLAMA_DIR]/LLaMA --model_size 7B --output_dir ./7B

  3. Modify hyperparams in finetune.py

    MICRO_BATCH_SIZE = 12
BATCH_SIZE = 36
EPOCHS = 3
LEARNING_RATE = 2e-5

  4. Run experiments

    python finetune.py