ros/LLaMA-Factory
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

modity code structure

Browse Source 
Former-commit-id: 0682ed357210897e0b67c4a6eb31a94b3eb929f1
This commit is contained in:
hiyouga
2023-07-15 16:54:28 +08:00
parent fa06b168ab
commit 6261fb362a
57 changed files with 1999 additions and 1816 deletions
Download Patch File Download Diff File Expand all files Collapse all files

19
README.md
View File

@@ -95,7 +95,7 @@ huggingface-cli login
- Python 3.8+ and PyTorch 1.13.1+
- 🤗Transformers, Datasets, Accelerate, PEFT and TRL
- jieba, rouge-chinese and nltk (used at evaluation)
- gradio and mdtex2html (used in web_demo.py)
- gradio and matplotlib (used in web_demo.py)
- uvicorn, fastapi and sse-starlette (used in api_demo.py)
And **powerful GPUs**!
@@ -137,7 +137,8 @@ python -m transformers.models.llama.convert_llama_weights_to_hf \
### (Continually) Pre-Training
```bash
CUDA_VISIBLE_DEVICES=0 python src/train_pt.py \
CUDA_VISIBLE_DEVICES=0 python src/train_bash.py \
--stage pt \
--model_name_or_path path_to_your_model \
--do_train \
--dataset wiki_demo \
@@ -158,7 +159,8 @@ CUDA_VISIBLE_DEVICES=0 python src/train_pt.py \
### Supervised Fine-Tuning
```bash
CUDA_VISIBLE_DEVICES=0 python src/train_sft.py \
CUDA_VISIBLE_DEVICES=0 python src/train_bash.py \
--stage sft \
--model_name_or_path path_to_your_model \
--do_train \
--dataset alpaca_gpt4_en \
@@ -179,7 +181,8 @@ CUDA_VISIBLE_DEVICES=0 python src/train_sft.py \
### Reward Model Training
```bash
CUDA_VISIBLE_DEVICES=0 python src/train_rm.py \
CUDA_VISIBLE_DEVICES=0 python src/train_bash.py \
--stage rm \
--model_name_or_path path_to_your_model \
--do_train \
--dataset comparison_gpt4_en \
@@ -199,7 +202,8 @@ CUDA_VISIBLE_DEVICES=0 python src/train_rm.py \
### PPO Training (RLHF)
```bash
CUDA_VISIBLE_DEVICES=0 python src/train_ppo.py \
CUDA_VISIBLE_DEVICES=0 python src/train_bash.py \
--stage ppo \
--model_name_or_path path_to_your_model \
--do_train \
--dataset alpaca_gpt4_en \
@@ -222,7 +226,7 @@ CUDA_VISIBLE_DEVICES=0 python src/train_ppo.py \
```bash
accelerate config # configure the environment
accelerate launch src/train_XX.py # arguments (same as above)
accelerate launch src/train_bash.py # arguments (same as above)
```
<details><summary>Example configuration for full-tuning with DeepSpeed ZeRO-2</summary>
@@ -256,7 +260,8 @@ use_cpu: false
### Evaluation (BLEU and ROUGE_CHINESE)
```bash
CUDA_VISIBLE_DEVICES=0 python src/train_sft.py \
CUDA_VISIBLE_DEVICES=0 python src/train_bash.py \
--stage pt \
--model_name_or_path path_to_your_model \
--do_eval \
--dataset alpaca_gpt4_en \