2026-10-11 16:37 UTC

post-training

band: coolmomentum: stable score: 0.088
temperature history

Episodes (4)

Independent deployments will determine whether LMSYS Miles v0.1 provides a reliable production-oriented stack that materially simplifies open-model post-training workflows.
expiredknownscott: medium
Independent training and evaluation will determine whether the released Scaffold CoT dataset improves reasoning accuracy and concision in sub-5B-parameter models over free-form chain-of-thought data.
expiredconvergesscott: medium
Surge AI claims post-training Qwen3.5-122B-A10B on non-coding office tasks improves SWE-Bench Pro by 5.8 percentage points, suggesting long-horizon workflow training can strengthen coding agents without software-specific training tasks.
seedconvergesscott: medium
Nathan Lambert and Tom Zick launch Trillium Labs, a nonprofit claiming it will publish fully open post-training recipes — data, code, evaluations, intermediate checkpoints — with backing from Halcyon Futures and Schmidt Sciences; whether it completes its coalition fundraising, ships its first open recipe, and attracts independent research use resolves whether an independent open-post-training institution can hold against frontier-lab secrecy.
seedconvergesscott: high

Trajectory notes