Archive training sources, portable recipes and provenance for reproducibility

reproduction/SHA256SUMS

@@ -0,0 +1,4 @@
+d72608eb7ce5ce3fced959544c7c5a869ff6ae574d2b9e87e2d7aa1ac0ac1f4a  chembl3d-explicit-h.tar.gz
+7add66a5db756b459e15c6c9d0bf75bf6110d007c17f2cc30fbcde4c58ef9695  chembl3d-implicit-h.tar.gz
+d61d4edc6c530d8efe03016316beecdb30cff1b8228f7c433bb4f98f59606695  geom-explicit-h.tar.gz
+06aa71e19be7423097ccb08540a463bb7f4fa6d99152428fff0cc47eb0ae8c8f  qm9-explicit-h.tar.gz

reproduction/chembl3d-explicit-h.tar.gz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d72608eb7ce5ce3fced959544c7c5a869ff6ae574d2b9e87e2d7aa1ac0ac1f4a
+size 422898

reproduction/chembl3d-implicit-h.tar.gz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7add66a5db756b459e15c6c9d0bf75bf6110d007c17f2cc30fbcde4c58ef9695
+size 241231

reproduction/geom-explicit-h.tar.gz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d61d4edc6c530d8efe03016316beecdb30cff1b8228f7c433bb4f98f59606695
+size 423904

reproduction/manifest.json

@@ -0,0 +1,120 @@
+{
+  "chembl3d-explicit-h": {
+    "model": "chembl3d-explicit-h",
+    "source_snapshot": "chembl3d_recovery_20260927_v2",
+    "published_completed_epochs": [
+      200,
+      300,
+      400
+    ],
+    "original_resume_completed_epoch": 5,
+    "original_validation_every_epochs": 5,
+    "reproduction_validation_every_epochs": 1,
+    "checkpoint_every_epochs": 5,
+    "seed": 42,
+    "devices": 2,
+    "gpu": "NVIDIA GeForce RTX 4090",
+    "batch_per_gpu": 320,
+    "accumulation": 1,
+    "precision": "bf16-mixed",
+    "recipe_changes": [
+      "Portable data/output paths",
+      "Fresh initialization; no historical resume",
+      "Epoch limit instead of allocation timer",
+      "W&B disabled",
+      "Evaluation every epoch; best and retained checkpoints every 5 epochs (QM9: 25)"
+    ],
+    "environment_note": "Python 3.13.7. Installed package versions captured from the shared training environment on 2026-10-08; not an original per-job environment lock.",
+    "limits": "Data and historical resume checkpoints are not distributed. Same recipe and seed do not imply a bitwise replay across resumes, hardware or software changes.",
+    "archive": "chembl3d-explicit-h.tar.gz",
+    "bytes": 422898,
+    "sha256": "d72608eb7ce5ce3fced959544c7c5a869ff6ae574d2b9e87e2d7aa1ac0ac1f4a"
+  },
+  "chembl3d-implicit-h": {
+    "model": "chembl3d-implicit-h",
+    "source_snapshot": "chembl3d_implicit_20260928_v2",
+    "published_completed_epochs": [
+      100,
+      200,
+      300
+    ],
+    "original_resume_completed_epoch": null,
+    "original_validation_every_epochs": 1,
+    "reproduction_validation_every_epochs": 1,
+    "checkpoint_every_epochs": 5,
+    "seed": 42,
+    "devices": 2,
+    "gpu": "NVIDIA GeForce RTX 4090",
+    "batch_per_gpu": 320,
+    "accumulation": 1,
+    "precision": "bf16-mixed",
+    "recipe_changes": [
+      "Portable data/output paths",
+      "Fresh initialization; no historical resume",
+      "Epoch limit instead of allocation timer",
+      "W&B disabled",
+      "Evaluation every epoch; best and retained checkpoints every 5 epochs (QM9: 25)"
+    ],
+    "environment_note": "Python 3.13.7. Installed package versions captured from the shared training environment on 2026-10-08; not an original per-job environment lock.",
+    "limits": "Data and historical resume checkpoints are not distributed. Same recipe and seed do not imply a bitwise replay across resumes, hardware or software changes.",
+    "archive": "chembl3d-implicit-h.tar.gz",
+    "bytes": 241231,
+    "sha256": "7add66a5db756b459e15c6c9d0bf75bf6110d007c17f2cc30fbcde4c58ef9695"
+  },
+  "geom-explicit-h": {
+    "model": "geom-explicit-h",
+    "source_snapshot": "geom_resume_e48_20261008",
+    "published_completed_epochs": [
+      50
+    ],
+    "original_resume_completed_epoch": 48,
+    "original_validation_every_epochs": 1,
+    "reproduction_validation_every_epochs": 1,
+    "checkpoint_every_epochs": 5,
+    "seed": 42,
+    "devices": 2,
+    "gpu": "NVIDIA GeForce RTX 4090",
+    "batch_per_gpu": 320,
+    "accumulation": 1,
+    "precision": "bf16-mixed",
+    "recipe_changes": [
+      "Portable data/output paths",
+      "Fresh initialization; no historical resume",
+      "Epoch limit instead of allocation timer",
+      "W&B disabled",
+      "Evaluation every epoch; best and retained checkpoints every 5 epochs (QM9: 25)"
+    ],
+    "environment_note": "Python 3.13.7. Installed package versions captured from the shared training environment on 2026-10-08; not an original per-job environment lock.",
+    "limits": "Data and historical resume checkpoints are not distributed. Same recipe and seed do not imply a bitwise replay across resumes, hardware or software changes.",
+    "archive": "geom-explicit-h.tar.gz",
+    "bytes": 423904,
+    "sha256": "d61d4edc6c530d8efe03016316beecdb30cff1b8228f7c433bb4f98f59606695"
+  },
+  "qm9-explicit-h": {
+    "model": "qm9-explicit-h",
+    "source_snapshot": "qm9_launch_20261008",
+    "published_completed_epochs": [],
+    "original_resume_completed_epoch": null,
+    "original_validation_every_epochs": 1,
+    "reproduction_validation_every_epochs": 1,
+    "checkpoint_every_epochs": 25,
+    "seed": 42,
+    "devices": 2,
+    "gpu": "NVIDIA GeForce RTX 4090",
+    "batch_per_gpu": 1024,
+    "accumulation": 2,
+    "precision": "bf16-mixed",
+    "recipe_changes": [
+      "Portable data/output paths",
+      "Fresh initialization; no historical resume",
+      "Epoch limit instead of allocation timer",
+      "W&B disabled",
+      "Evaluation every epoch; best and retained checkpoints every 5 epochs (QM9: 25)"
+    ],
+    "environment_note": "Python 3.13.7. Installed package versions captured from the shared training environment on 2026-10-08; not an original per-job environment lock.",
+    "limits": "Data and historical resume checkpoints are not distributed. Same recipe and seed do not imply a bitwise replay across resumes, hardware or software changes.",
+    "archive": "qm9-explicit-h.tar.gz",
+    "bytes": 423543,
+    "sha256": "06aa71e19be7423097ccb08540a463bb7f4fa6d99152428fff0cc47eb0ae8c8f"
+  }
+}

reproduction/qm9-explicit-h.tar.gz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:06aa71e19be7423097ccb08540a463bb7f4fa6d99152428fff0cc47eb0ae8c8f
+size 423543

reproduction/validation.json

@@ -0,0 +1,22 @@
+{
+  "chembl3d-explicit-h": {
+    "status": "passed",
+    "files_verified": 142,
+    "check": "Portable train.sh --cfg job on two ranks; data metadata checks only; no fitting"
+  },
+  "chembl3d-implicit-h": {
+    "status": "passed",
+    "files_verified": 141,
+    "check": "Portable train.sh --cfg job on two ranks; data metadata checks only; no fitting"
+  },
+  "geom-explicit-h": {
+    "status": "passed",
+    "files_verified": 150,
+    "check": "Portable train.sh --cfg job on two ranks; data metadata checks only; no fitting"
+  },
+  "qm9-explicit-h": {
+    "status": "passed",
+    "files_verified": 150,
+    "check": "Portable train.sh --cfg job on two ranks; data metadata checks only; no fitting"
+  }
+}