Lab polish: CQL alpha-trajectory + Dreamer policy cleanup#19
Merged
Conversation
Final follow-up commit for the two long-program reproduction labs spawned in Wave E: - labs/rl_decision/lab_cql_offline_minigrid/: added the alpha-tuning trajectory plot (ablation_alpha_traj.png) showing how the dual variable adapts; notebook now writes it from the auto-tune branch. - labs/world_models/lab_dreamer_cartpole_pixels/: policy.py polish (cleanup of imagination-rollout reward bookkeeping that the trainer agent left mid-edit before the conversation handed off). Both labs remain end-to-end runnable. https://claude.ai/code/session_017Ez7KNKDCGRRLjEnJi9TW7
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Closing touch on the CQL lab and a Dreamer policy polish.
labs/rl_decision/lab_cql_offline_minigrid/:assets/ablation_alpha_traj.pngshowing the dual variable α adapting (starts at 1.0, slides to 0.6 because the empirical CQL gap ~1.3 is below the target=5).q_overestimation.png: DQN's Q_OOD climbs above the optimal return while Q_seen stays put — the textbook offline-RL pathology. CQL's gap (Q_OOD − Q_seen) stays at zero and grows negative, demonstrating the lower-bound property.labs/world_models/lab_dreamer_cartpole_pixels/: small policy cleanup for the imagination-rollout reward bookkeeping.https://claude.ai/code/session_017Ez7KNKDCGRRLjEnJi9TW7
Generated by Claude Code