Read on Twitter
当今开源 LLM 模型的快照，每个构建者都应该知道的 🧵 
-- 商业用途 --
1. 𝗙𝗹𝗮𝗻-𝗨𝗟𝟮 Google’s open source LLM huggingface.co/google/flan-ul2
2. 𝗢𝗽𝗲𝗻𝗖𝗵𝗮𝘁𝗞𝗶𝘁 Create specialized and general purpose chatbots huggingface.co/spaces/togethe…
1. 𝗙𝗹𝗮𝗻-𝗨𝗟𝟮 Google’s open source LLM huggingface.co/google/flan-ul2
2. 𝗢𝗽𝗲𝗻𝗖𝗵𝗮𝘁𝗞𝗶𝘁 Create specialized and general purpose chatbots huggingface.co/spaces/togethe…
3. 𝗖𝗲𝗿𝗲𝗯𝗿𝗮𝘀-𝗚𝗣𝗧 A family of seven GPT-3 models from 111M to 13B parameters huggingface.co/cerebras/Cereb……
4. 𝗣𝘆𝘁𝗵𝗶𝗮 A family of 16 language models from 70M-12B parameters from EleutherAI huggingface.co/EleutherAI/pyt…
4. 𝗣𝘆𝘁𝗵𝗶𝗮 A family of 16 language models from 70M-12B parameters from EleutherAI huggingface.co/EleutherAI/pyt…
5. 𝗕𝗹𝗼𝗼𝗺 & 𝗺𝗧𝗢 Models capable of following human instructions in dozens of languages zero-shot huggingface.co/bigscience/blo……
6. 𝗻𝗮𝗻𝗼𝗧𝟱 Pre-training and fine-tuning T5-style models github.com/PiotrNawrot/na……
7. 𝗚𝗲𝗼𝗩 A 9B pre-trained LLM github.com/geov-ai/geov
6. 𝗻𝗮𝗻𝗼𝗧𝟱 Pre-training and fine-tuning T5-style models github.com/PiotrNawrot/na……
7. 𝗚𝗲𝗼𝗩 A 9B pre-trained LLM github.com/geov-ai/geov
--研究用途--
1. 𝗕𝗮𝗶𝘇𝗲 Open-source chat model trained with LoRA using 100k dialogs generated by letting ChatGPT chat with itself huggingface.co/spaces/project……
2. 𝗩𝗶𝗰𝘂𝗻𝗮 Open-source chatbot achieving almost the same performance as ChatGPT
chat.lmsys.org
1. 𝗕𝗮𝗶𝘇𝗲 Open-source chat model trained with LoRA using 100k dialogs generated by letting ChatGPT chat with itself huggingface.co/spaces/project……
2. 𝗩𝗶𝗰𝘂𝗻𝗮 Open-source chatbot achieving almost the same performance as ChatGPT
chat.lmsys.org
3. 𝗞𝗼𝗮𝗹𝗮 Chatbot trained by fine-tuning Meta’s LLaMA on dialogue data gathered from the web chat.lmsys.org/?model=koala-1……
4. 𝗚𝗣𝗧𝟰𝗔𝗹𝗹 Train an assistant-style LLM with ~800k GPT-3.5-Turbo Generations based on LLaMa github.com/nomic-ai/gpt4a…
4. 𝗚𝗣𝗧𝟰𝗔𝗹𝗹 Train an assistant-style LLM with ~800k GPT-3.5-Turbo Generations based on LLaMa github.com/nomic-ai/gpt4a…
5. 𝗟𝗶𝘁-𝗟𝗟𝗮𝗠𝗔 Independent implementation of LLaMA that builds on nanoGPT github.com/Lightning-AI/l……
6. 𝗗𝗼𝗹𝗹𝘆 (𝗗𝗮𝘁𝗮𝗯𝗿𝗶𝗰𝗸𝘀) An LLM trained using GPT-J and fine tuned on Stanford Alpaca Code github.com/databrickslabs…
6. 𝗗𝗼𝗹𝗹𝘆 (𝗗𝗮𝘁𝗮𝗯𝗿𝗶𝗰𝗸𝘀) An LLM trained using GPT-J and fine tuned on Stanford Alpaca Code github.com/databrickslabs…
7. 𝗗𝗮𝗹𝗮𝗶 The fastest way to run LLaMA and Alpaca locally (includes a user interface) github.com/cocktailpeanut……
8. 𝗖𝗼𝗹𝗼𝘀𝘀𝗮𝗹𝗖𝗵𝗮𝘁 LLM trained with RLHF powered by Colossal-AI github.com/hpcaitech/Colo…
8. 𝗖𝗼𝗹𝗼𝘀𝘀𝗮𝗹𝗖𝗵𝗮𝘁 LLM trained with RLHF powered by Colossal-AI github.com/hpcaitech/Colo…
8. 𝗔𝗹𝗽𝗮𝗰𝗮.𝗰𝗽𝗽 Run a fast ChatGPT-like model locally on your device github.com/antimatter15/a……
9. 𝗔𝗹𝗽𝗮𝗰𝗮-𝗟𝗼𝗥𝗔 An Instruct model of similar quality to text-davinci-003 that can run on local devices huggingface.co/spaces/tloen/a…
9. 𝗔𝗹𝗽𝗮𝗰𝗮-𝗟𝗼𝗥𝗔 An Instruct model of similar quality to text-davinci-003 that can run on local devices huggingface.co/spaces/tloen/a…
10. 𝗹𝗹𝗮𝗺𝗮.𝗰𝗽𝗽 Inference of LLaMA model in pure C/C++ supporting LLaMA, Alpaca, GPT4All, and Vicuna Code github.com/ggerganov/llam…
原作者：@theaievangelist 4月11日发布
我希望你发现这个线程有帮助。
如果您这样做了，我将不胜感激：
1.关注我 @FinanceYF5 获取更多内容。
2. Like/Retweet 下面的推文分享给他人：
我希望你发现这个线程有帮助。
如果您这样做了，我将不胜感激：
1.关注我 @FinanceYF5 获取更多内容。
2. Like/Retweet 下面的推文分享给他人：
https://twitter.com/FinanceYF5/status/1661693000660189184
• • •
Missing some Tweet in this thread? You can try to
force a refresh
