Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/juncongmoo/chatllama
ChatLLaMA ๐ข Open source implementation for LLaMA-based ChatGPT runnable in a single GPU. 15x faster training process than ChatGPT
https://github.com/juncongmoo/chatllama
Last synced: 5 days ago
JSON representation
ChatLLaMA ๐ข Open source implementation for LLaMA-based ChatGPT runnable in a single GPU. 15x faster training process than ChatGPT
- Host: GitHub
- URL: https://github.com/juncongmoo/chatllama
- Owner: juncongmoo
- Created: 2023-02-27T19:08:45.000Z (almost 2 years ago)
- Default Branch: main
- Last Pushed: 2023-03-07T09:07:26.000Z (almost 2 years ago)
- Last Synced: 2024-11-21T06:36:41.346Z (21 days ago)
- Language: Python
- Homepage:
- Size: 28.3 KB
- Stars: 1,207
- Watchers: 20
- Forks: 138
- Open Issues: 5
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
- awesome-open-chatgpt - juncongmoo/chatllama - based ChatGPT runnable in a single GPU. ([TavernAI/TavernAI](https://github.com/TavernAI/TavernAI) / ๆฐๆฎ)
- awesome-totally-open-chatgpt - juncongmoo/chatllama - based ChatGPT runnable in a single GPU. ([tatsu-lab/stanford_alpaca](https://github.com/tatsu-lab/stanford_alpaca) / Other LLaMA-derived projects:)
- awesome-chatgpt - juncongmoo/chatllama - ChatLLaMA is an open-source implementation for LLaMA-based ChatGPT that can be run on a single GPU. It offers a training process that is 15 times faster than ChatGPT. (Other / Other sdk/libraries)
README
# ChatLLaMA
> ๐ข Open source implementation for LLaMA-based ChatGPT runnable in a single GPU. 15x faster training process than `ChatGPT`
- ๐ฅ Please check [`pyllama`](https://github.com/juncongmoo/pyllama) for `LLaMA` installation and `single GPU inference` setup.
- ๐ฅ To train ChatGPT in 5 mins - [minichatgpt](https://github.com/juncongmoo/minichatgpt)
Meta has recently released LLaMA, a collection of foundational large language models ranging from 7 to 65 billion parameters.
LLaMA is creating a lot of excitement because it is smaller than GPT-3 but has better performance. For example, LLaMA's 13B architecture outperforms GPT-3 despite being 10 times smaller. This new collection of fundamental models opens the door to faster inference performance and chatGPT-like real-time assistants, while being cost-effective and running on a single GPU.
However, LLaMA was not fine-tuned for instruction task with a Reinforcement Learning from Human Feedback (RLHF) training process.
The good news is that we introduce `ChatLLaMA`, the first open source implementation of LLaMA based on RLHF:
- A complete open source implementation that enables you to build a ChatGPT-style service based on pre-trained LLaMA models.
- Compared to the original ChatGPT, the training process and single-GPU inference are much faster and cheaper by taking advantage of the smaller size of LLaMA architectures.
- ChatLLaMA has built-in support for DeepSpeed ZERO to speedup the fine-tuning process.
- The library also supports all LLaMA model architectures (7B, 13B, 33B, 65B), so that you can fine-tune the model according to your preferences for training time and inference performance.
Image from [OpenAIโs blog](https://openai.com/blog/chatgpt).
# Installation
```
pip install chatllama
```
# Get started with ChatLLaMA
> :warning: Please note this code represents the algorithmic implementation for RLHF training process of LLaMA and does not contain the model weights. To access the model weights, you need to apply to Meta's [form](https://forms.gle/jk851eBVbX1m5TAv5).
ChatLLaMA allows you to easily train LLaMA-based architectures in a similar way to ChatGPT, using RLHF.
For example, below is the code to start the training in the case of ChatLLaMA 7B.
```python
from chatllama.rlhf.trainer import RLTrainer
from chatllama.rlhf.config import Config
path = "path_to_config_file.yaml"
config = Config(path=path)
trainer = RLTrainer(config.trainer)
trainer.distillate()
trainer.train()
trainer.training_stats.plot()
```
Note that you should provide Meta's original weights and your custom dataset before starting the fine-tuning process. Alternatively, you can generate your own dataset using LangChain's agents.
```python
python generate_dataset.py
```
> The code is originally from [nebuly-ai](https://github.com/nebuly-ai/nebullvm/tree/main/apps/accelerate/chatllama) with some changes. More changes will follow up soon. And the original license link is [here](https://github.com/nebuly-ai/nebullvm/blob/main/LICENSE).