Inside LFM2.5-2.6B's post-training recipe: SFT, RL, multi-domain distillation
helloiamleonie · x · 2026-09-25
Leonie assembles the full post-training recipe for LFM2.5-2.6B: (1) classical SFT with agent traces and antidoom training; (2) SFT+RL to train specialized teachers per skill; (3) MOPD distills all skills back into one checkpoint; (4) agentic RL polishes the model for agent harnesses. The 2.6B model competes with the 3x larger Qwen3.5-9B.
More from Models
- Micius-Flash: 27B Multimodal Model for Hardware and Semiconductor Agents — zhiman-ai · 2026-11-11
- Baidu says RSI auto-optimized ERNIE inference, 6.24x decode speedup on Kunxin P800 — xiaohu · 2026-11-11
- Baidu's ECHO: multi-token prediction reusing the last layer as a drafter — xiaohu · 2026-11-11
- ERNIE 5.5 Flash Preview: 280B-param multi-router MoE activating only 14B per token — xiaohu · 2026-11-11
- ERNIE 5.5 recreates the Pelican test: 117-second animation drawn frame-by-frame in Python — xiaohu · 2026-11-11
- 35,000+ Grok-built apps exposed patient data, plaintext passwords, and sales pipeline — The Decoder · 2026-11-11