sim-square-broad-r02-baseline-idql

Part of Mulligan. Browse evaluations at Policy Arena; the datasets are in the mulligan organization.

State-based IDQL agent: diffusion actor with scalar IQL critic (policy.pt, a PyTorch checkpoint, and stats.json normalizers).

Property	Value
Task	sim-square-broad
Model round	R2
Arm	baseline
Campaign cell	square_broad_r2_baseline_uniform_no_cf_human_only
Seeds	1, 2, 3, 4, 5 (one folder per seed)
Training step	250001

Checkpoints

Folder	Training run config
seed-1	release/run-configs/sim-square-broad-r02-baseline-idql__seed-1.json
seed-2	release/run-configs/sim-square-broad-r02-baseline-idql__seed-2.json
seed-3	release/run-configs/sim-square-broad-r02-baseline-idql__seed-3.json
seed-4	release/run-configs/sim-square-broad-r02-baseline-idql__seed-4.json
seed-5	release/run-configs/sim-square-broad-r02-baseline-idql__seed-5.json

Evaluations

Held-out initial-state grid; per-rollout outcomes are in the linked dataset.

Dataset	Folder	N	Successes
sim-square-broad-r00-r03-eval	seed-1	1	21551/30000
sim-square-broad-r00-r03-eval	seed-2	1	22282/30000
sim-square-broad-r00-r03-eval	seed-3	1	21455/30000
sim-square-broad-r00-r03-eval	seed-4	1	20150/30000
sim-square-broad-r00-r03-eval	seed-5	1	21170/30000

Data

Trained on: sim-square-broad-c00-teleop-baseline, sim-square-broad-c01-baseline-policy-rollouts, sim-square-broad-c01-dagger-baseline, sim-square-broad-c02-baseline-policy-rollouts, sim-square-broad-c02-dagger-baseline.

Referenced by the metadata of: sim-square-broad-r00-r03-eval.

Provenance

SHA-256 of every file is recorded in release.json. .pt files are PyTorch pickles; load them only in a trusted environment. The run configs are in the Mulligan code repository; the recipes there retrain each checkpoint. License: mit.