Upload MiniHack ReMDM planner checkpoints

README.md

@@ -25,7 +25,7 @@ Code, configs and evaluation harness: https://github.com/mathisweil/minihack-ReM
 | Path | Role | Architecture | Params | Trained to | Full state |
 |---|---|---|---|---|---|
 | `checkpoints/offline/Minihack-OfflineDiffusion-BC-82M` | Diffusion planner (offline BC) | 4L, d_model 256, 4 heads, horizon 64 | 5.2M | 40,000 gradient steps, 81,920,000 sample-equivalents | 80 MB |
-| `checkpoints/online/Minihack-OnlineDiffusion-DAgger-123M` | Diffusion planner (online DAgger) | 4L, d_model 256, 4 heads, horizon 64 | 5.2M | DAgger iteration 600 | 80 MB |
+| `checkpoints/online/Minihack-Online-Diffusion-DAgger-100M` | Diffusion planner (online DAgger) | 4L, d_model 256, 4 heads, horizon 64 | 5.2M | DAgger iteration 563, 5,654,965 env steps | 80 MB |
 
 Each directory holds three things: the original `.pth` training state (weights,
 EMA shadow, optimiser, scheduler, and for the DAgger run the curriculum and RNG
@@ -56,11 +56,11 @@ repository.
 from huggingface_hub import snapshot_download
 
 # everything
-snapshot_download(repo_id="MathisW78/remdm-minihack-checkpoints", local_dir=".")
+snapshot_download(repo_id="mathisweil/remdm-minihack-checkpoints", local_dir=".")
 
 # inference weights only
 snapshot_download(
-    repo_id="MathisW78/remdm-minihack-checkpoints",
+    repo_id="mathisweil/remdm-minihack-checkpoints",
     local_dir=".",
     allow_patterns=["**/model.safetensors", "**/config*.yaml"],
 )

checkpoints/offline/Minihack-OfflineDiffusion-BC-82M/model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:75b037d8368fd413ae0d2c16541f7b3dce2adadce21a39c11175c1c391a3f9da
+oid sha256:62f95eb01da61cd1a436f1dc07ec44a57b7624ce209d1be51e51b17642045eae
 size 20974876

checkpoints/online/Minihack-Online-Diffusion-DAgger-100M/config.yaml

@@ -0,0 +1,92 @@
+action_dim: 12
+aux_loss_weight: 0.5
+baselines_bc_batch_size: 256
+baselines_bc_epochs: 50
+baselines_bc_lr: 0.0003
+baselines_bc_oracle_episodes_per_env: 5000
+baselines_dqn_buffer_size: 100000
+baselines_dt_batch_size: 256
+baselines_dt_context_len: 64
+baselines_dt_embed_dim: 256
+baselines_dt_epochs: 50
+baselines_dt_eval_max_steps: 200
+baselines_dt_lr: 0.0003
+baselines_dt_max_ep_len: 200
+baselines_dt_n_heads: 4
+baselines_dt_n_layers: 4
+baselines_dt_oracle_episodes_per_env: 5000
+baselines_eval_episodes_per_env: null
+baselines_eval_freq_env_steps: 10000
+baselines_n_envs_per_id: 2
+baselines_output_dir: outputs/baselines
+baselines_wandb_project: minihack-ReMDM-baselines
+buffer_capacity: 10000
+checkpoint_dir: online_s0/dagger_20260811_210823_ee39
+checkpoint_eval_episodes: 50
+checkpoint_every_timesteps: 940000
+collect_episodes_per_env: 5000
+collect_num_workers: 32
+collect_output: data/oracle_bc_qmul.pt
+crop_size: 9
+curriculum_preseed: true
+curriculum_queue_size: 100
+dagger_batch_size: 2048
+dagger_grad_clip: 1.0
+dagger_lr: 3.0e-05
+device: cuda
+diffusion_steps_collect: 5
+diffusion_steps_eval: 10
+dropout: 0.0
+efficiency_multiplier: 1.5
+ema_decay: 0.999
+episodes_per_iteration: 30
+eta: 0.15
+eval_episodes_per_env: 50
+global_gate_init: -3.0
+grad_steps_per_iteration: 100
+id_envs:
+- MiniHack-Room-Random-5x5-v0
+- MiniHack-Room-Random-15x15-v0
+- MiniHack-Corridor-R2-v0
+- MiniHack-MazeWalk-9x9-v0
+id_eval_every_timesteps: 470000
+label_smoothing: 0.0
+loss_weight_clip: 1000.0
+map_h: 21
+map_w: 79
+mask_token: 12
+n_embd: 256
+n_global_tokens: 8
+n_head: 4
+n_layer: 4
+noise_schedule: linear
+num_collection_workers: 32
+num_diffusion_steps: 100
+offline_batch_size: 2048
+offline_buffer_capacity: 1500000
+offline_checkpoint_every_grad_steps: 10000
+offline_eval_every_grad_steps: 5000
+offline_grad_clip: 1.0
+offline_log_every: 50
+offline_lr: 0.0003
+offline_total_grad_steps: 60000
+ood_envs:
+- MiniHack-Room-Dark-15x15-v0
+- MiniHack-Corridor-R5-v0
+- MiniHack-MazeWalk-45x19-v0
+ood_eval_every_timesteps: 470000
+pad_token: 13
+physics_aware_sampling: false
+remask_strategy: conf
+replan_every: 16
+save_policy: true
+seed: 0
+seq_len: 64
+temperature: 0.5
+top_p: 0.9
+torch_compile: true
+total_timesteps: 5650000
+use_amp: true
+use_global_stream: true
+use_wandb: true
+weight_decay: 0.0001

checkpoints/online/Minihack-Online-Diffusion-DAgger-100M/iter563.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:04d10a4cd119dfe8273bfb06a57ba0e8aaaefa5816a7825f62581dad28144676
+size 83995501

checkpoints/online/Minihack-Online-Diffusion-DAgger-100M/model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1a25616b8ebd7673982fea78ba33030a552cebd0356cbe9a6ff55c42380a3709
+size 20974876

checkpoints/online/Minihack-Online-Diffusion-DAgger-100M/selection.json

@@ -0,0 +1,27 @@
+{
+  "policy": "best-of-N over periodic checkpoints",
+  "selected": {
+    "dagger_iteration": 563
+  },
+  "selection_metric": null,
+  "candidates": {
+    "unit": "dagger_iterations",
+    "every": null,
+    "configured_max": null
+  },
+  "eval_protocol": {
+    "episodes_per_env": 50,
+    "weights": "ema",
+    "id_envs": [
+      "MiniHack-Room-Random-5x5-v0",
+      "MiniHack-Room-Random-15x15-v0",
+      "MiniHack-Corridor-R2-v0",
+      "MiniHack-MazeWalk-9x9-v0"
+    ],
+    "ood_envs": [
+      "MiniHack-Room-Dark-15x15-v0",
+      "MiniHack-Corridor-R5-v0",
+      "MiniHack-MazeWalk-45x19-v0"
+    ]
+  }
+}