开源大语言模型(LLM)汇总
随着ChatGPT的火爆，越来越多人希望在本地运行一个大语言模型。为此我维护了这个开源大语言模型汇总，跟踪每天不发的大语言模型和精调语言模型。
我将根据个模型采用的基础大模型进行分类，每个大模型下列出各派生模型。
文章目录
Alpaca (Stanford)
- 🏠Alpaca 官网: https://crfm.stanford.edu/2023/03/13/alpaca.html
- ❤️Alpaca GitHub: https://github.com/tatsu-lab/stanford_alpaca
- 💵是否可以商用: 否
以下是基于 Stanford Alpaca 项目的衍生模型或类似模型：
- Alpaca.cpp
- Alpaca-LoRA
- Baize
- Cabrita
- BELLE
- Luotuo
- Vicuna
- Chinese-Vicuna
- GPT4All
- Koala
- llama.cpp
- Lit-LLaMA ️
Alpaca.cpp
Alpaca-LoRA
Baize
Cabrita
- ❤️GitHub: https://github.com/22-hours/cabrita
BELLE
- ❤️GitHub: https://github.com/LianjiaTech/BELLE
Luotuo
- ❤️GitHub: https://github.com/LC1332/Luotuo-Chinese-LLM
- ❤️GitHub: https://github.com/LC1332/Chinese-alpaca-lora
Vicuna (FastChat)
- ❤️GitHub: https://github.com/lm-sys/FastChat
- 🎬视频: Vicuna — 90% of ChatGPT quality by using a new dataset? — YouTube
Chinese-Vicuna
- ❤️GitHub: https://github.com/Facico/Chinese-Vicuna
GPT4All
Koala
- 📖博客: Koala: A Dialogue Model for Academic Research — The Berkeley Artificial Intelligence Research Blog
- ❤️GitHub: EasyLM/koala.md at main · young-geng/EasyLM (github.com)
- 💻Demo: FastChat (lmsys.org)
- 🎬视频: Investigating Koala a ChatGPT style Dialogue Model — YouTube
llama.cpp
- ❤️GitHub: https://github.com/ggerganov/llama.cpp
Lit-LLaMA ️
BLOOM (BigScience)
- ❤️Hugging Face: bigscience/bloom · Hugging Face
- 💻Hugging Face Demo: Bloom Demo — a Hugging Face Space by huggingface
以下是基于 BigScience BLOOM 项目的衍生模型或类似模型：
- BLOOM-LoRA
- Petals
BLOOM-LoRA
Petals
Flamingo (Google/Deepmind)
以下是基于 Flamingo 项目的衍生模型或类似模型：
- Flamingo — Pytorch
- OpenFlamingo
Flamingo — Pytorch
- ❤️GitHub: https://github.com/lucidrains/flamingo-pytorch
OpenFlamingo
FLAN (Google)
- ❤️GitHub: https://github.com/google-research/FLAN
以下是基于 FLAN 项目的衍生模型或类似模型：
- Flan-Alpaca
- Flan-UL2
Flan-Alpaca
Flan-UL2
- ❤️Hugging Face: google/flan-ul2 · Hugging Face
- 🎬视频: Trying Out Flan 20B with UL2 — Working in Colab with 8Bit Inference — YouTube
GLM (General Language Model)
以下是基于 GLM 项目的衍生模型或类似模型：
- GLM-130B
- ChatGLM-6B
GLM-130B
- ❤️GitHub: https://github.com/THUDM/GLM-130B
ChatGLM-6B
- 🏠官网: ChatGLM
- ❤️GitHub: https://github.com/THUDM/ChatGLM-6B
GPT-J (EleutherAI)
- ❤️GitHub: https://github.com/kingoflolz/mesh-transformer-jax/#gpt-j-6b
- 💻Demo: https://6b.eleuther.ai/
以下是基于 GPT-J 项目的衍生模型或类似模型：
- Dolly
Dolly (Databricks)
Cerebras-GPT (Cerebras)
- 🏠官网: Cerebras-GPT: A Family of Open, Compute-efficient, Large Language Models — Cerebras
- ❤️Hugging Face: cerebras (Cerebras) (huggingface.co)
- 🎬视频: Checking out the Cerebras-GPT family of models — YouTube
GPT-NeoX
- ❤️GitHub: https://github.com/EleutherAI/gpt-neox
HuggingGPT
- ❤️GitHub: https://github.com/microsoft/JARVIS
- 📑论文: https://arxiv.org/abs/2303.17580
Polyglot
- ❤️GitHub: https://github.com/EleutherAI/polyglot
Pythia
- ❤️GitHub: https://github.com/EleutherAI/pythia
Segment Anything
- 🏠官网: Introducing Segment Anything: Working toward the first foundation model for image segmentation (facebook.com)
- ❤️GitHub: https://github.com/facebookresearch/segment-anything
The RWKV Language Model
- ❤️GitHub: https://github.com/BlinkDL/RWKV-LM/RWKV-LM
- ❤️ChatRWKV: https://github.com/BlinkDL/ChatRWKV
- 💻Hugging Face Demo: HuggingFace Gradio demo (14B ctx8192)
- 💻Hugging Face Demo: Raven (7B finetuned on Alpaca) Demo
- 🎬视频: Raven — RWKV-7B RNN’s LLM Strikes Back — YouTube
XGLM
- ❤️GitHub: https://github.com/facebookresearch/fairseq/tree/main/examples/xglm
- ❤️Hugging Face: https://huggingface.co/docs/transformers/model_doc/xglm