Today we're launching Miles v0.1, an open-source RL framework for LLMs and multimodal models.
RL training is easy to start and hard to debug. Miles helps you ensure your run is correct, use hardware efficiently, and keep RL running at scale. Over the past 9 months, 72
Regular evening in the ai frenzy
1. asked Codex to install Qwen-3.8 27B on 8gb gpu laptop 2. asked Qwen to build minesweeper (failed) 3. asked Codex to steer Qwen to build minesweeper (success after 2 iterations in 2h) 4. played minesweeper, set 14 sec as baseline on easy 5.
Wanted To See PEAK Tok/s on SGLang with Qwen 3.8 27B running w/ DSPARK using 0xSero Recipe.
[NVLINK] 🚀 Qwen3.8-27B running on Qwen3.8-27B · SGLang + DSpark in PROD 🖥️ Dual 3090 GPUs (GPU0 + GPU1) — running 1 parallel coding agent 📊 The Numbers: • Total: 2,891 tokens in 9s
You have reached the end of the archive
All of qwen38