Strona główna Odkrywaj Pomoc
Zaloguj się
dongzenan
/
alpaca-lora
1
0
Forkuj 0
Pliki Problemy 0 Oczekujące zmiany 0 Wiki

Brak opisu

Drzewo: 63121244c8
Gałęzie Tagi
alpaca-lora
Eric Wang 63121244c8 Licenses and whatnot 3 lat temu
.gitignore 26f64780ad initial commit 3 lat temu
DATA_LICENSE 63121244c8 Licenses and whatnot 3 lat temu
LICENSE 63121244c8 Licenses and whatnot 3 lat temu
README.md 63121244c8 Licenses and whatnot 3 lat temu
alpaca_data.json 26f64780ad initial commit 3 lat temu
conversion.py 26f64780ad initial commit 3 lat temu
finetune.py 63121244c8 Licenses and whatnot 3 lat temu
generate.py 63121244c8 Licenses and whatnot 3 lat temu
iteration.ipynb 26f64780ad initial commit 3 lat temu
lengths.ipynb 26f64780ad initial commit 3 lat temu

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