Add CodeLeWM checkpoint artifact codelewm-action-use-20260520-6650183

checkpoints/codelewm-action-use-20260520-6650183/checkpoints/checkpoint.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1e361498c722893c9754abcc9c2efa4499a615590572b77c7f0de939e789ac66
+size 351495773

checkpoints/codelewm-action-use-20260520-6650183/checkpoints/checkpoint.pt.manifest.json

@@ -0,0 +1,14 @@
+{
+  "checkpoint_path": "checkpoint.pt",
+  "checkpoint_sha256": "1e361498c722893c9754abcc9c2efa4499a615590572b77c7f0de939e789ac66",
+  "metadata": {
+    "action_view": "text",
+    "config_hash": "7903d135afcbc66a8c3a55859d67fbceea2ef9bbe4623bec9250ba31e8409de9",
+    "latent_dim": 256,
+    "model_class": "TorchCodeTransitionModel",
+    "record_schema_version": "codelewm.transition.v1",
+    "schema_version": "codelewm.checkpoint.v1"
+  },
+  "migration_hook": null,
+  "schema_version": "codelewm.checkpoint.v1"
+}

checkpoints/codelewm-action-use-20260520-6650183/config.json

@@ -0,0 +1,54 @@
+{
+  "data": {
+    "manifest": "/CodeLeWM/.artifacts/hf/codelewm-action-use-20260520-6650183/pack/manifest.json",
+    "train": "/CodeLeWM/.artifacts/hf/codelewm-action-use-20260520-6650183/pack/hdf5/train.hdf5",
+    "val": "/CodeLeWM/.artifacts/hf/codelewm-action-use-20260520-6650183/pack/hdf5/val.hdf5"
+  },
+  "loader": {
+    "batch_size": 64,
+    "num_workers": 4,
+    "persistent_workers": true,
+    "pin_memory": true,
+    "shuffle": true
+  },
+  "loss": {
+    "action_use_margin": 0.02,
+    "action_use_margin_weight": 0.25,
+    "enable_action_use_margin": true,
+    "enable_retrieval_loss": false,
+    "retrieval_temperature": 0.1,
+    "retrieval_weight": 0.0,
+    "sigreg_knots": 17,
+    "sigreg_num_proj": 1024,
+    "sigreg_weight": 0.09
+  },
+  "name": "codelewm_scaled_action_use_margin_gpu_a10g",
+  "optimizer": {
+    "lr": 0.0001,
+    "type": "AdamW",
+    "weight_decay": 0.001
+  },
+  "output": {
+    "checkpoint_dir": "/CodeLeWM/.artifacts/hf/codelewm-action-use-20260520-6650183/train/checkpoints",
+    "manifest_path": "/CodeLeWM/.artifacts/hf/codelewm-action-use-20260520-6650183/train/training_manifest.json",
+    "metrics_path": "/CodeLeWM/.artifacts/hf/codelewm-action-use-20260520-6650183/train/metrics.jsonl",
+    "run_dir": "/CodeLeWM/.artifacts/hf/codelewm-action-use-20260520-6650183/train"
+  },
+  "schema_version": "codelewm.train_config.v1",
+  "seed": 240119,
+  "trainer": {
+    "accelerator": "gpu",
+    "devices": 1,
+    "gradient_clip_val": 1.0,
+    "max_steps": 60000,
+    "precision": "bf16-mixed"
+  },
+  "wm": {
+    "action_sequence_length": 256,
+    "action_view": "text",
+    "embed_dim": 256,
+    "history_size": 1,
+    "num_preds": 1,
+    "state_sequence_length": 1024
+  }
+}

checkpoints/codelewm-action-use-20260520-6650183/manifest.json

@@ -0,0 +1,111 @@
+{
+  "artifact_id": "training_run-ce98fe8768af2143",
+  "artifact_kind": "training_run",
+  "command": [
+    "codelewm",
+    "train",
+    "--config",
+    "/CodeLeWM/.artifacts/hf/codelewm-action-use-20260520-6650183/configs/train.json",
+    "--executor",
+    "torch",
+    "--device",
+    "auto",
+    "--overwrite",
+    "--json",
+    "--log-jsonl",
+    "/CodeLeWM/.artifacts/hf/codelewm-action-use-20260520-6650183/logs/train.jsonl"
+  ],
+  "config_sha256": "a950dab9a3511f531f2afda08edef0d291f5cc7f6747826e95caa6a2cea8b039",
+  "created_at": "2026-05-20T10:16:48Z",
+  "files": [
+    {
+      "bytes": 1709,
+      "path": "config.json",
+      "sha256": "508212142c2e7add15c11200b96fc983fce3656f41e443a9ac25b9554491845e"
+    },
+    {
+      "bytes": 1282,
+      "path": "metrics.jsonl",
+      "sha256": "b06e0fdc2a427b9e82d9e5481242b1e814adcce56adfc0bd6b718dc0c52bd8b8"
+    },
+    {
+      "bytes": 1408,
+      "path": "reports/metrics_report.json",
+      "sha256": "1c7cae88ef9521f3363e0e66773bc8b67921a1b5253ab157ca1d9582755d4cd8"
+    },
+    {
+      "bytes": 351495773,
+      "path": "checkpoints/checkpoint.pt",
+      "sha256": "1e361498c722893c9754abcc9c2efa4499a615590572b77c7f0de939e789ac66"
+    },
+    {
+      "bytes": 511,
+      "path": "checkpoints/checkpoint.pt.manifest.json",
+      "sha256": "29be4cea4092bb8e514215b3d486545be8423e0e6c9797efca3d225e5745c512"
+    },
+    {
+      "bytes": 2461,
+      "path": "reports/torch_training_report.json",
+      "sha256": "46b142008a953d322a73f11c59ca4d5338fcf214527713b3e4c55e66aca28823"
+    }
+  ],
+  "metadata": {
+    "dataset_manifest_path": ".artifacts/hf/codelewm-action-use-20260520-6650183/pack/manifest.json",
+    "executor": {
+      "checkpoint_schema_version": "codelewm.checkpoint.v1",
+      "device": "cuda",
+      "executor": "torch",
+      "objective": {
+        "action_use_margin": 0.02,
+        "action_use_margin_weight": 0.25,
+        "enable_action_use_margin": true,
+        "enable_retrieval_loss": false,
+        "retrieval_temperature": 0.1,
+        "retrieval_weight": 0.0,
+        "sigreg_knots": 17,
+        "sigreg_num_proj": 1024,
+        "sigreg_weight": 0.09
+      },
+      "precision": "bf16-mixed",
+      "torch": "2.12.0+cu130",
+      "train_rows": 18019,
+      "val_rows": 1291
+    },
+    "final_metrics": {
+      "collapse/effective_rank": 5.913853737656158,
+      "collapse/effective_rank_ratio": 0.023100991162719367,
+      "collapse/embedding_count": 192.0,
+      "collapse/embedding_norm_mean": 14.052607765565478,
+      "collapse/latent_dim": 256.0,
+      "collapse/nearest_neighbor_entropy": 4.850434265952372,
+      "collapse/pairwise_cosine_mean": 0.007218573232229116,
+      "collapse/per_dim_variance_max": 3.4118030796654626,
+      "collapse/per_dim_variance_median": 0.643422042029191,
+      "collapse/per_dim_variance_min": 0.09125475415772162,
+      "loss/action_use_margin": 0.006401113234460354,
+      "loss/action_use_margin_weighted": 0.0016002783086150885,
+      "loss/prediction_mse": 0.005575100891292095,
+      "loss/sigreg": 1.03125,
+      "loss/sigreg_weighted": 0.0927734375,
+      "loss/total": 0.09994881600141525,
+      "train/examples": 3833852.0,
+      "train/examples_per_second": 1088.7440435079823,
+      "train/gradient_norm": 1.5844553709030151,
+      "val/loss/action_use_margin": 0.05521763453171367,
+      "val/loss/action_use_margin_weighted": 0.013804408632928417,
+      "val/loss/prediction_mse": 0.18526783337195715,
+      "val/loss/sigreg": 2.2153775095939636,
+      "val/loss/sigreg_weighted": 0.19938398400942484,
+      "val/loss/total": 0.39845622721172513
+    },
+    "run_id": "codelewm_scaled_action_use_margin_gpu_a10g",
+    "schema_version": "codelewm.training_run.v1",
+    "seed": 240119,
+    "step_count": 60000
+  },
+  "parent_artifacts": [
+    "dataset-67895f8dc3e217c4"
+  ],
+  "schema_version": "codelewm.artifact_manifest.v1",
+  "source_git_sha": "6650183eda75ecced9f826d5f2875548808c39f0"
+}

checkpoints/codelewm-action-use-20260520-6650183/metrics.jsonl

@@ -0,0 +1 @@
+{"metrics": {"collapse/effective_rank": 5.913853737656158, "collapse/effective_rank_ratio": 0.023100991162719367, "collapse/embedding_count": 192.0, "collapse/embedding_norm_mean": 14.052607765565478, "collapse/latent_dim": 256.0, "collapse/nearest_neighbor_entropy": 4.850434265952372, "collapse/pairwise_cosine_mean": 0.007218573232229116, "collapse/per_dim_variance_max": 3.4118030796654626, "collapse/per_dim_variance_median": 0.643422042029191, "collapse/per_dim_variance_min": 0.09125475415772162, "loss/action_use_margin": 0.006401113234460354, "loss/action_use_margin_weighted": 0.0016002783086150885, "loss/prediction_mse": 0.005575100891292095, "loss/sigreg": 1.03125, "loss/sigreg_weighted": 0.0927734375, "loss/total": 0.09994881600141525, "train/examples": 3833852.0, "train/examples_per_second": 1088.7440435079823, "train/gradient_norm": 1.5844553709030151, "val/loss/action_use_margin": 0.05521763453171367, "val/loss/action_use_margin_weighted": 0.013804408632928417, "val/loss/prediction_mse": 0.18526783337195715, "val/loss/sigreg": 2.2153775095939636, "val/loss/sigreg_weighted": 0.19938398400942484, "val/loss/total": 0.39845622721172513}, "run_id": "codelewm_scaled_action_use_margin_gpu_a10g", "schema_version": "codelewm.training_metrics.v1", "step": 60000}

checkpoints/codelewm-action-use-20260520-6650183/reports/metrics_report.json

@@ -0,0 +1,32 @@
+{
+  "final_metrics": {
+    "collapse/effective_rank": 5.913853737656158,
+    "collapse/effective_rank_ratio": 0.023100991162719367,
+    "collapse/embedding_count": 192.0,
+    "collapse/embedding_norm_mean": 14.052607765565478,
+    "collapse/latent_dim": 256.0,
+    "collapse/nearest_neighbor_entropy": 4.850434265952372,
+    "collapse/pairwise_cosine_mean": 0.007218573232229116,
+    "collapse/per_dim_variance_max": 3.4118030796654626,
+    "collapse/per_dim_variance_median": 0.643422042029191,
+    "collapse/per_dim_variance_min": 0.09125475415772162,
+    "loss/action_use_margin": 0.006401113234460354,
+    "loss/action_use_margin_weighted": 0.0016002783086150885,
+    "loss/prediction_mse": 0.005575100891292095,
+    "loss/sigreg": 1.03125,
+    "loss/sigreg_weighted": 0.0927734375,
+    "loss/total": 0.09994881600141525,
+    "train/examples": 3833852.0,
+    "train/examples_per_second": 1088.7440435079823,
+    "train/gradient_norm": 1.5844553709030151,
+    "val/loss/action_use_margin": 0.05521763453171367,
+    "val/loss/action_use_margin_weighted": 0.013804408632928417,
+    "val/loss/prediction_mse": 0.18526783337195715,
+    "val/loss/sigreg": 2.2153775095939636,
+    "val/loss/sigreg_weighted": 0.19938398400942484,
+    "val/loss/total": 0.39845622721172513
+  },
+  "run_id": "codelewm_scaled_action_use_margin_gpu_a10g",
+  "schema_version": "codelewm.training_metrics.v1",
+  "step_count": 60000
+}

checkpoints/codelewm-action-use-20260520-6650183/reports/torch_training_report.json

@@ -0,0 +1,67 @@
+{
+  "collapse_report": {
+    "effective_rank": 5.913853737656158,
+    "effective_rank_ratio": 0.023100991162719367,
+    "embedding_count": 192,
+    "embedding_norm_mean": 14.052607765565478,
+    "latent_dim": 256,
+    "nearest_neighbor_entropy": 4.850434265952372,
+    "pairwise_cosine_mean": 0.007218573232229116,
+    "per_dim_variance_max": 3.4118030796654626,
+    "per_dim_variance_median": 0.643422042029191,
+    "per_dim_variance_min": 0.09125475415772162,
+    "schema_version": "codelewm.eval.collapse_report.v1"
+  },
+  "dataset": {
+    "action_view": "text",
+    "train_rows": 18019,
+    "val_rows": 1291
+  },
+  "metrics": {
+    "collapse/effective_rank": 5.913853737656158,
+    "collapse/effective_rank_ratio": 0.023100991162719367,
+    "collapse/embedding_count": 192.0,
+    "collapse/embedding_norm_mean": 14.052607765565478,
+    "collapse/latent_dim": 256.0,
+    "collapse/nearest_neighbor_entropy": 4.850434265952372,
+    "collapse/pairwise_cosine_mean": 0.007218573232229116,
+    "collapse/per_dim_variance_max": 3.4118030796654626,
+    "collapse/per_dim_variance_median": 0.643422042029191,
+    "collapse/per_dim_variance_min": 0.09125475415772162,
+    "loss/action_use_margin": 0.006401113234460354,
+    "loss/action_use_margin_weighted": 0.0016002783086150885,
+    "loss/prediction_mse": 0.005575100891292095,
+    "loss/sigreg": 1.03125,
+    "loss/sigreg_weighted": 0.0927734375,
+    "loss/total": 0.09994881600141525,
+    "train/examples": 3833852.0,
+    "train/examples_per_second": 1088.7440435079823,
+    "train/gradient_norm": 1.5844553709030151,
+    "val/loss/action_use_margin": 0.05521763453171367,
+    "val/loss/action_use_margin_weighted": 0.013804408632928417,
+    "val/loss/prediction_mse": 0.18526783337195715,
+    "val/loss/sigreg": 2.2153775095939636,
+    "val/loss/sigreg_weighted": 0.19938398400942484,
+    "val/loss/total": 0.39845622721172513
+  },
+  "objective": {
+    "action_use_margin": 0.02,
+    "action_use_margin_weight": 0.25,
+    "enable_action_use_margin": true,
+    "enable_retrieval_loss": false,
+    "retrieval_temperature": 0.1,
+    "retrieval_weight": 0.0,
+    "sigreg_knots": 17,
+    "sigreg_num_proj": 1024,
+    "sigreg_weight": 0.09
+  },
+  "run_id": "codelewm_scaled_action_use_margin_gpu_a10g",
+  "runtime": {
+    "device": "cuda",
+    "dtype": "torch.bfloat16",
+    "precision": "bf16-mixed",
+    "torch": "2.12.0+cu130"
+  },
+  "schema_version": "codelewm.torch_training_report.v1",
+  "step_count": 60000
+}

checkpoints/codelewm-action-use-20260520-6650183/training_manifest.json

@@ -0,0 +1,84 @@
+{
+  "artifact_manifest_id": "training_run-ce98fe8768af2143",
+  "artifact_manifest_path": "manifest.json",
+  "checkpoint_files": [
+    {
+      "bytes": 351495773,
+      "path": "checkpoints/checkpoint.pt",
+      "sha256": "1e361498c722893c9754abcc9c2efa4499a615590572b77c7f0de939e789ac66"
+    },
+    {
+      "bytes": 511,
+      "path": "checkpoints/checkpoint.pt.manifest.json",
+      "sha256": "29be4cea4092bb8e514215b3d486545be8423e0e6c9797efca3d225e5745c512"
+    }
+  ],
+  "config_path": "config.json",
+  "config_sha256": "a950dab9a3511f531f2afda08edef0d291f5cc7f6747826e95caa6a2cea8b039",
+  "dataset_manifest_path": ".artifacts/hf/codelewm-action-use-20260520-6650183/pack/manifest.json",
+  "final_metrics": {
+    "collapse/effective_rank": 5.913853737656158,
+    "collapse/effective_rank_ratio": 0.023100991162719367,
+    "collapse/embedding_count": 192.0,
+    "collapse/embedding_norm_mean": 14.052607765565478,
+    "collapse/latent_dim": 256.0,
+    "collapse/nearest_neighbor_entropy": 4.850434265952372,
+    "collapse/pairwise_cosine_mean": 0.007218573232229116,
+    "collapse/per_dim_variance_max": 3.4118030796654626,
+    "collapse/per_dim_variance_median": 0.643422042029191,
+    "collapse/per_dim_variance_min": 0.09125475415772162,
+    "loss/action_use_margin": 0.006401113234460354,
+    "loss/action_use_margin_weighted": 0.0016002783086150885,
+    "loss/prediction_mse": 0.005575100891292095,
+    "loss/sigreg": 1.03125,
+    "loss/sigreg_weighted": 0.0927734375,
+    "loss/total": 0.09994881600141525,
+    "train/examples": 3833852.0,
+    "train/examples_per_second": 1088.7440435079823,
+    "train/gradient_norm": 1.5844553709030151,
+    "val/loss/action_use_margin": 0.05521763453171367,
+    "val/loss/action_use_margin_weighted": 0.013804408632928417,
+    "val/loss/prediction_mse": 0.18526783337195715,
+    "val/loss/sigreg": 2.2153775095939636,
+    "val/loss/sigreg_weighted": 0.19938398400942484,
+    "val/loss/total": 0.39845622721172513
+  },
+  "metadata": {
+    "executor": {
+      "checkpoint_schema_version": "codelewm.checkpoint.v1",
+      "device": "cuda",
+      "executor": "torch",
+      "objective": {
+        "action_use_margin": 0.02,
+        "action_use_margin_weight": 0.25,
+        "enable_action_use_margin": true,
+        "enable_retrieval_loss": false,
+        "retrieval_temperature": 0.1,
+        "retrieval_weight": 0.0,
+        "sigreg_knots": 17,
+        "sigreg_num_proj": 1024,
+        "sigreg_weight": 0.09
+      },
+      "precision": "bf16-mixed",
+      "torch": "2.12.0+cu130",
+      "train_rows": 18019,
+      "val_rows": 1291
+    }
+  },
+  "metrics_path": "metrics.jsonl",
+  "metrics_report_path": "reports/metrics_report.json",
+  "parent_artifacts": [
+    "dataset-67895f8dc3e217c4"
+  ],
+  "report_files": [
+    {
+      "bytes": 2461,
+      "path": "reports/torch_training_report.json",
+      "sha256": "46b142008a953d322a73f11c59ca4d5338fcf214527713b3e4c55e66aca28823"
+    }
+  ],
+  "run_id": "codelewm_scaled_action_use_margin_gpu_a10g",
+  "schema_version": "codelewm.training_run.v1",
+  "seed": 240119,
+  "step_count": 60000
+}