0.2.0 — playable WM toolkit for DeepSeek Harness

Install

dsh plugin --profile wm add github:WayneJin0918/dsh-wm
dsh --profile wm

Or play sunset with no GPU / no Harness:

node cli.js inspect fixtures/sunset --indices first,mid,last
node cli.js diagnose "is Sora a world simulator"
node cli.js knowledge --id wm-routes

Pitch

听说梁文锋觉得世界模型没什么用？更贴近闭门会报道的说法是：3D / 视频生成 / 世界模型不在 DeepSeek 的 AGI 主线上——「商业上是好生意，跟智能路线图没有什么关系」。

公司可以不做产品主线。Harness 是开源研究操作系统。DSH-WM 给它加上一块 WM 实验室：看条带、认三条路线（3D 显示 / Pixel Video Gen / Latent Prediction）、给 run 打分、对 skill 和 wm.yaml 做 RSI。

Liang’s reported take is that world models are off DeepSeek’s AGI mainline. The harness is still the place researchers work. This plugin is the WM lab on top of it.

In this release

• Measure: wm_discover, wm_summarize, wm_rollout_diff, wm_inspect
• Know: three-route map + technique cards (wm_knowledge, wm_diagnose)
• Iterate: wm-run-triage, wm-knowledge, wm-rsi, wm-ablation, wm-revisit
• Playground: fixtures/sunset (late-horizon wipe, no cluster)

Repo: https://github.com/WayneJin0918/dsh-wm