Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/ssbuild/visualglm_finetuning
https://github.com/ssbuild/visualglm_finetuning
qlora visualglm
Last synced: 4 days ago
JSON representation
- Host: GitHub
- URL: https://github.com/ssbuild/visualglm_finetuning
- Owner: ssbuild
- License: apache-2.0
- Created: 2023-10-08T15:57:48.000Z (about 1 year ago)
- Default Branch: dev
- Last Pushed: 2024-04-23T16:42:37.000Z (7 months ago)
- Last Synced: 2024-04-28T04:59:07.993Z (7 months ago)
- Topics: qlora, visualglm
- Language: Python
- Homepage:
- Size: 6.86 MB
- Stars: 3
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.MD
- License: LICENSE
Awesome Lists containing this project
README
## statement
- [deep_training](https://github.com/ssbuild/deep_training)
```text
2024-04-22 简化
2023-10-18 微调推理测试初步完成
2023-10-17 initial visualglm_finetuning
```
## install
- pip install -U -r requirements.txt
- 如果无法安装 , 可以切换官方源 pip install -i https://pypi.org/simple -U -r requirements.txt
## weight
- [visualglm-6b](https://huggingface.co/THUDM/visualglm-6b)
## data sample
open_data https://github.com/ssbuild/open_data
单条数据示例
```text
p prefix optional
q question optional
a answer must
```
```json
{"id": 1, "paragraph": [{"q": "![]()
../assets/demo.jpeg\n图中的狗是什么品种？", "a": "图中是一只拉布拉多犬。"}]}
```
或者
```json
{
"id": 0,
"conversations": [
{
"from": "user",
"value": "![]()
../assets/demo.jpeg\n图中的狗是什么品种？"
},
{
"from": "assistant",
"value": "图中是一只拉布拉多犬。"
}
]
}
```
## infer
# infer.py 推理预训练模型
# infer_finetuning.py 推理微调模型
# infer_lora_finetuning.py 推理lora微调模型
python infer.py
| **量化等级** | **最低 GPU 显存** |
| -------------- | ----------------- |
| FP16（无量化） | 13 GB |
| INT8 | 10 GB |
| INT4 | 6 GB |
## training
```text
# 制作数据
cd scripts
bash train_full.sh -m dataset
or
bash train_lora.sh -m dataset
or
bash train_ptv2.sh -m dataset
注: num_process_worker 为多进程制作数据 ， 如果数据量较大 ， 适当调大至cpu数量
dataHelper.make_dataset_with_args(data_args.train_file,mixed_data=False, shuffle=True,mode='train',num_process_worker=0)
# 全参数训练
bash train_full.sh -m train
# lora adalora ia3
bash train_lora.sh -m train
# ptv2
bash train_ptv2.sh -m train
```
## 训练参数
[训练参数](args.MD)
## 友情链接
- [pytorch-task-example](https://github.com/ssbuild/pytorch-task-example)
- [chatmoss_finetuning](https://github.com/ssbuild/chatmoss_finetuning)
- [chatglm_finetuning](https://github.com/ssbuild/chatglm_finetuning)
- [chatglm2_finetuning](https://github.com/ssbuild/chatglm2_finetuning)
- [t5_finetuning](https://github.com/ssbuild/t5_finetuning)
- [llm_finetuning](https://github.com/ssbuild/llm_finetuning)
- [llm_rlhf](https://github.com/ssbuild/llm_rlhf)
- [chatglm_rlhf](https://github.com/ssbuild/chatglm_rlhf)
- [t5_rlhf](https://github.com/ssbuild/t5_rlhf)
- [rwkv_finetuning](https://github.com/ssbuild/rwkv_finetuning)
- [baichuan_finetuning](https://github.com/ssbuild/baichuan_finetuning)
- [baichuan2_finetuning](https://github.com/ssbuild/baichuan_finetuning)
- [xverse_finetuning](https://github.com/ssbuild/xverse_finetuning)
- [aigc_serving](https://github.com/ssbuild/aigc_serving)
- [aigc_evals](https://github.com/ssbuild/aigc_evals)
##
纯粹而干净的代码
## Reference
https://github.com/THUDM/VisualGLM-6B
## Star History
[](https://star-history.com/#ssbuild/visualglm_finetuning&Date)