Wave E batch 3: Dreamer-pixels and CQL-offline reproduction labs#17
Merged
Conversation
labs/world_models/lab_dreamer_cartpole_pixels/
- Dreamer-style world model on CartPole-v1 from pixels.
- CNN encoder + RSSM (det h_t + stochastic Gaussian z_t) + transposed-
conv decoder + reward head; latent imagination actor-critic with
GAE lambda-returns.
- README, paper.md (linking paper_world_models / paper_dreamer_v2 /
paper_dreamer_v3), notebook narrative, full src module split
(env, world_model, trainer, policy, viz, seeds). Training/asset
PNGs are deferred to a follow-up; the code is end-to-end runnable.
labs/rl_decision/lab_cql_offline_minigrid/
- CQL on an 8x8 sparse-reward gridworld, with BC + DQN baselines and
an alpha-tuning ablation.
- Three trainers (BC, DQN, CQL) plus a unified eval pipeline.
- assets/ contains the four story PNGs: q_overestimation,
q_overestimation_dqn_only, ood_action_density, action_histogram,
eval_returns, ablation_alpha.
- data/ checkpoints: bc.pt, dqn.pt, cql.pt, offline_dataset.pt.
- paper.md links paper_cql / paper_bear / paper_iql.
Both labs follow the per-lab directory contract documented in
labs/RESTRUCTURE_PROPOSAL.md.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Two more long-program reproduction labs in the new paradigm-grouped layout.
labs/world_models/lab_dreamer_cartpole_pixels/— Dreamer-style world model on CartPole-v1 from pixels:paper_world_models/paper_dreamer_v2/paper_dreamer_v3), notebook narrative, fullsrc/module split.labs/rl_decision/lab_cql_offline_minigrid/— CQL vs BC vs DQN on an 8×8 sparse-reward gridworld:assets/:q_overestimation,q_overestimation_dqn_only,ood_action_density,action_histogram,eval_returns,ablation_alpha.data/:bc.pt,dqn.pt,cql.pt,offline_dataset.pt.paper_cql,paper_bear,paper_iql.Both labs follow the per-lab directory contract documented in
labs/RESTRUCTURE_PROPOSAL.md.https://claude.ai/code/session_017Ez7KNKDCGRRLjEnJi9TW7
Generated by Claude Code