bigscience-workshop / petals

🌸 Run LLMs at home, BitTorrent-style. Fine-tuning and inference up to 10x faster than offloading
https://petals.dev
MIT License
8.98k stars 498 forks source link

petals prompt-tuning server node no response #270

Open roy-shih opened 1 year ago

roy-shih commented 1 year ago

Hello

I want to learn and contribute to this great work, so I run the example to learn more about it, but when I run prompt-tuning , I had a problem with the server not responding. Have I missed something?

截圖 2023-02-25 上午3 24 58

BTW, is there possible to publish a more detailed tutorial on how to create a private group in the future? For example, create some tutorials on youtube.

thank you so much!

best

borzunov commented 1 year ago

Hi @roy-shih,

We have just released v1.1.3 with multiple bug fixes, could you please try running prompt tuning again? If it doesn't help, please share the full logs (the error message is cut off on the screenshot).

Recording a video tutorial for setting up a private swarm is a great idea! I'd love to do that once I'm done with current high-priority tasks (incentive system, leaderboard, some bug fixes).