Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/ModelTC/EasyLLM
Built upon Megatron-Deepspeed and HuggingFace Trainer, EasyLLM has reorganized the code logic with a focus on usability. While enhancing usability, it also ensures training efficiency.
https://github.com/ModelTC/EasyLLM
Last synced: about 1 month ago
JSON representation
Built upon Megatron-Deepspeed and HuggingFace Trainer, EasyLLM has reorganized the code logic with a focus on usability. While enhancing usability, it also ensures training efficiency.
- Host: GitHub
- URL: https://github.com/ModelTC/EasyLLM
- Owner: ModelTC
- License: apache-2.0
- Created: 2023-11-26T10:12:52.000Z (10 months ago)
- Default Branch: main
- Last Pushed: 2024-07-25T08:34:06.000Z (about 2 months ago)
- Last Synced: 2024-07-29T00:36:57.218Z (about 2 months ago)
- Language: Python
- Size: 970 KB
- Stars: 32
- Watchers: 8
- Forks: 4
- Open Issues: 1
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
- Awesome-LLM-Compression - [Code
README
# EasyLLM
Built upon Megatron-Deepspeed and HuggingFace Trainer, EasyLLM has reorganized the code logic with a focus on usability. While enhancing usability, it also ensures training efficiency.
## Install
- Install python requirements
```shell
pip install -r requirements.txt
```
other dependency
* flash-attn (dropout_layer_norm) (maybe you need to compile it by yourself)
- Pull deepspeed & add them to pythonpath
```shell
export PYTHONPATH=/path/to/DeepSpeed:$PYTHONPATH
```
- Install package in development mode
```shell
pip install -e . -v
```
## Train
[Train Example](./docs/train.md)
## Infer and Eval
[Infer Example](./docs/infer.md)
## Support Models
* qwen14b,
* internlm7-20b,
* baichuan1/2 (7b-13b)
* llama1-2 (7b/13b/70b)
[Model Example](./docs/model.md)
## Data
[Data Example](./docs/data.md)
## 3D Parallel config setting
[Parallel Example](docs/parallel.md)
## Speed Benchmark
[Speed Benchmark](docs/benchmark.md)
## Dynamic Checkpoint
To optimize the model training performance in terms of time and space, EasyLLM supports Dynamic Checkpoint. Based on the input token size, it enables checkpointing for some layers. The configuration file settings are as follows:
[Dynamic Checkpoint Example](docs/dc.md)
## License
This repository is released under the [Apache-2.0](LICENSE) license.
## Acknowledgement
We learned a lot from the following projects when developing EasyLLM.
- [DeepSpeed](https://github.com/microsoft/DeepSpeed.git)
- [Megatron-DeepSpeed](https://github.com/microsoft/Megatron-DeepSpeed.git)
- [Megatron-LM](https://github.com/NVIDIA/Megatron-LM.git)
- [Flash Attention 1&2](https://github.com/Dao-AILab/flash-attention)
- [LightLLM](https://github.com/ModelTC/lightllm)
- [Huggingface Transformers](https://github.com/huggingface/transformers.git)