Article
Qwen3.8-27B-pi: Effort-Ordered Reasoning for Agentic Coding
bytkim
• • 13Thomas Kim PRO
bytkim
AI & ML interests
None yet
Recent Activity
new activity 1 day ago
bytkim/Qwen3.8-27B-pi-GGUF:dflash2 new activity 1 day ago
bytkim/Qwen3.8-27B-pi:Pls nvfp4 ver. repliedto their post 1 day ago
Introducing Qwen3.8-27B-pi! Fine-tuned for the Pi coding harness to turn plans into working, checked implementations.
Trained on curated Pi coding sessions, then refined through reinforcement learning to balance task success with more economical reasoning at low and medium effort, while keeping xhigh focused on correctness.
79.78% on Terminal-Bench 2.1, versus Base model's 75.28%, with approximately 12% fewer output tokens.
Available in BF16, FP8, and GGUF, with serving guides and MTP/DFlash2 companions!
https://huggingface.co/collections/bytkim/qwen38-pi
https://huggingface.co/bytkim/Qwen3.8-27B-pi-FP8
https://huggingface.co/bytkim/Qwen3.8-27B-pi
https://huggingface.co/bytkim/Qwen3.8-27B-pi-GGUF
Organizations
None yet
published an article 2 days ago
published an article 4 months ago
Article
QLORA SFT Distillation Effects on Qwen3.6 27B Agentic Coding Harness Fluency
bytkim
• • 10