Agent RL for three models cost $1,177 total; LFM3 gained 43% tool efficiency
yb2698 · x · 2026-11-09
Key metrics on a 150-task held-out test set: all models showed tool efficiency gains, up to 43% for LFM3. Except for TwIL-8B pi and the LFM3 models, accuracy changes stayed within noise — likely because the set is still too easy for bigger models, though the signal is clear for LFM3. On the small (40-task) traj2env held-out set, most models were also within noise.
Total cost of the three training jobs on HF Jobs was about $1,177: LFM2.5 ran 36h for $720, TwIL-8B 13.5h for $271, LFM3 18.6h for $186. Sandbox costs excluded; cpu-basic sandbox instances run $0.01/hour.
More from coding & agent
- WonderDuck: open-source robot duck ships full sim-to-real pipeline — philfung · 2026-11-09
- AI Engineer Code Summit kicks off in SF with agent-heavy agenda — threepointone · 2026-11-09
- Agent Security Incidents Need Tamper-Proof Audit Trails, Khala Research Report Argues — 0xSammy · 2026-11-09
- SenseTime open-senses Code2Video: an open model that makes editable videos via code — HeyZoyaKhan · 2026-11-09
- Alexandr Wang: nobody writes code anymore — orchestrating agents is the job — 2C_ornot2C · 2026-11-09
- How a cloud AI agent joined a home network via Tailscale and drove a radio — Dave_Maynor · 2026-11-09