axion 1.0 trains a 35B agent model with just two light SFT passes and RL
petrusenko_max · x · 2026-11-04
axion 1.0 builds a 35B agent model with a minimal three-stage recipe: two lightweight SFT passes followed by RL, with no mid-training, DPO, or OPD. If the results hold, those commonly assumed stages stop being mandatory and competition shifts to data quality — likely decided by the evals.
More from Models
- AI semiotics: models are swayed by visual symbols without recognizing it in their chains-of-thought — emollick · 2026-11-05
- vLLM ships day-0 support for Kimi K3.1 across NVIDIA, AMD and TPUs — vllm_project · 2026-11-05
- Kimi K3.1 Pro hits 61.2% on long-horizon code bench, up from 55.95% — vllm_project · 2026-11-05
- Liquid AI: vision encoders can burn more compute than the LLM on large images and video — JosephJacks_ · 2026-11-05
- Baseten explains how to make flash models faster: kernel fusion, PDL, and host loop cuts — baseten · 2026-11-05
- Moonshot AI open-sources Kimi K3.1 Pro: 2.8T-param MoE with 1M-token context — Kimi_Moonshot · 2026-11-05