Awesome LLM
Large language models are essential for future lifes, but there are so many variants, we don't know which one is better as there is no really a right metrics to eval it. The best way it compare them one by one, choose the best performance one suitable for your hardware.
A collection of latest awesome GPTs. GPTs have flooded into our work life like a torrent. Generative Pretrained Models has brought evolutionary progress on many areas. We made this list available for tracking && researching on this exciting topic. Any PRs are welcomed!
LLMs
Ⅰ. ChatGPT LLM
The world most advancest AI system ever
- llama: https://github.com/facebookresearch/llama
- Vicuna: https://github.com/lm-sys/FastChat
- Chinese Vicuna: https://github.com/lm-sys/FastChat
- PandaLM: https://github.com/dandelionsllm/pandallm , looks like they finetuned on more precise data
Ⅱ. GPT4 (Multi-Modality)
LLM based image reasoning, more step forward compare to LLMs
- LLaVA: https://llava.hliu.cc/
- MiniGPT4: https://github.com/Vision-CAIR/MiniGPT-4
CodeGen
Papers on how to generate code
- CodeGen2: https://github.com/salesforce/CodeGen2
Typical Papers
papers you should read about LLM
- What Language Model Architecture and Pretraining Objective Work Best for Zero-Shot Generalization?
- CodeGen2: Lessons for Training LLMs on Programming and Natural Languages
LLM Evaluation
LLM Deployment
llama.cpp or MLC
- MLC-Chat: https://github.com/mlc-ai/mlc-llm/ IMO, this is faster than llama.cpp
- llama.cpp: Pure C++ solution for LLMs.
LLM Distillation
make it smaller
- LaMini-LM: https://github.com/mbzuai-nlp/LaMini-LM
LLM SpeedUp
Engineering on LLMs
- DeepSpeed:
- Colossal-AI:
- JAX:
Keep updating...