Growing, publicly downloadable video-backed rollout dataset for the trace ablation arm at training step 500, seed 20260921. Each selected frozen observational prefix has exactly one cached continuation in each of two routes: null-intervention observational and diagnostic do(X=1). Prefix IDs are sorted once within each (W,Z) stratum and remain balanced as the cohort grows. The direct route of an observationally trained arm is a diagnostic; it does not replace the paper's observational importance-weighted estimator. Sampling note. Rollouts use a growing key-value cache: every generated frame attends to the whole history since the prefix, up to 280 frames, at absolute temporal positions. The…
Independent publisher
Robert Osazuwa Ness
osazuwa
causal machine learning, probabilistic modeling, generative modeling
Models in Library0
Datasets in Library2
Models on Hugging Face11
Followers—
Compare the published evidence
This profile connects the 2 model and dataset records shown below to their upstream cards and sources. The library counts describe SAVRN's coverage, rather than this publisher's entire output.
Compare model specifications, source dates, and hardware estimates on the individual records. Publisher identity verification is separate from a model's quality, deployment readiness, or suitability for your work.
Datasets
Growing, publicly downloadable video-backed rollout dataset for the tracemarkov ablation arm at training step 500, seed 20260921. Each selected frozen observational prefix has exactly one cached continuation in each of two routes: null-intervention observational and diagnostic do(X=1). Prefix IDs are sorted once within each (W,Z) stratum and remain balanced as the cohort grows. The direct route of an observationally trained arm is a diagnostic; it does not replace the paper's observational importance-weighted estimator. Sampling note. Rollouts use a growing key-value cache: every generated frame attends to the whole history since the prefix, up to 280 frames, at absolute temporal positions.…