Wing Lian
@winglian.bsky.social
Founder and maintainer of Axolotl AI. ⚽🎿
The sandbox uses WebAssembly + Python multiprocessing to safely execute model-generated code in parallel, fully locally. This enables scalable, automated reward signals for GRPO fine-tuning without the complexity of Docker or remote eval infra.
Blog Post 🤗: hf.co/blog/axolotl...
🧵(2/3)
Blog Post 🤗: hf.co/blog/axolotl...
🧵(2/3)
Training Large Language Models with Interpreter Feedback using WebAssembly
A Blog post by Axolotl AI on Hugging Face
hf.co
April 3, 2025 at 4:48 PM
The sandbox uses WebAssembly + Python multiprocessing to safely execute model-generated code in parallel, fully locally. This enables scalable, automated reward signals for GRPO fine-tuning without the complexity of Docker or remote eval infra.
Blog Post 🤗: hf.co/blog/axolotl...
🧵(2/3)
Blog Post 🤗: hf.co/blog/axolotl...
🧵(2/3)
Needs backwards kernels so we can use it for finetuning 😅
November 20, 2024 at 4:46 PM
Needs backwards kernels so we can use it for finetuning 😅