File size: 908 Bytes
32dfe89
 
ea9357c
 
 
 
 
32dfe89
ea9357c
b19f02f
 
ea9357c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
cb9f38f
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
---
license: apache-2.0
datasets:
- TigerResearch/pretrain_zh
language:
- zh
pipeline_tag: text-generation
---

based [gpt2](https://huggingface.co/gpt2)

trained with [LLaMA-Efficient-Tuning](https://github.com/hiyouga/LLaMA-Efficient-Tuning)

```sh
python src/train_bash.py \
    --stage pt \
    --model_name_or_path gpt2 \
    --do_train \
    --finetuning_type full \
    --dataset tiger_pretrain_zh \
    --template ziya \
    --use_fast_tokenizer \
    --preprocessing_num_workers 64 \
    --per_device_train_batch_size 8 \
    --gradient_accumulation_steps 16 \
    --output_dir .cache/gpt2-tigerResearch_pretrain_zh \
    --lr_scheduler_type cosine \
    --logging_steps 10 \
    --save_steps 1000 \
    --eval_steps 500 \
    --learning_rate 5e-5 \
    --num_train_epochs 1.0 \
    --val_size 3000 \
    --evaluation_strategy steps \
    --plot_loss \
    --max_source_length 1024 \
    --bf16
```