OptimalScale/LMFlow

• Gold I

Code and documentation to train Stanford's Alpaca models, and generate the data.

• 1,610−6 Elo, change today
• 30.2k stars
• 3.9k forks
• 2y ago

git clone https://github.com/tatsu-lab/stanford_alpaca.git

• deep-learning
• instruction-following
• language-model

• Copper II

Google AI 2018 BERT pytorch implementation

• 1,337−1 Elo, change today
• 6.5k stars
• 1.3k forks
• 3y ago

git clone https://github.com/codertimo/BERT-pytorch.git

• bert
• language-model
• nlp
• pytorch

• Emerald V

A framework for few-shot evaluation of language models.

• 1,765+9 Elo, change today
• 14.4k stars
• 3.8k forks
• 53 watchers
• 11d ago

pipx install lm_eval

• evaluation-framework
• language-model
• transformer

• Gold II

An open source implementation of CLIP.

• 1,585−2 Elo, change today
• 14.2k stars
• 1.3k forks
• 5d ago

git clone https://github.com/mlfoundations/open_clip.git

• computer-vision
• contrastive-loss
• deep-learning
• language-model

• Emerald III

Nano vLLM

• 1,816−7 Elo, change today
• 16.4k stars
• 2.8k forks
• 86 watchers
• 6mo ago

git clone https://github.com/GeeeekExplorer/nano-vllm.git

• deep-learning
• inference
• llm
• nlp

• Silver III

RWKV (pronounced RwaKuv) is an RNN with great LLM performance, which can also be directly trained like a GPT transformer (parallelizable). We are at RWKV-7 "Goose". So it's combining the best of RNN and transformer - great performance, linear time, constant space (no kv-cache), fast training, infinite ctx_len, and free sentence embedding.

• 1,475−6 Elo, change today
• 14.7k stars
• 1k forks
• 11d ago

git clone https://github.com/BlinkDL/RWKV-LM.git

• attention-mechanism
• chatgpt
• deep-learning
• gpt