Add PyTorch golden reference dumps

ref/vjepa2-ac-vitg/frame0.context.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6bddc62812f7ddb6447bb8da6b02b9e920e510ce495642f3719cf80da03af045
+size 1441920

ref/vjepa2-ac-vitg/frame0.frames_u8.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a6e6474b5ddb62e96b666ea91d2e0dd887109d236d3e5d757add9c8390a7da51
+size 393344

ref/vjepa2-ac-vitg/frame0.input.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cf0bd5f847af1d50752728ce0408973c2c4651917adcf216ebcf135c87cf30b7
+size 1572992

ref/vjepa2-ac-vitg/frame0.last_hidden_state.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f8d17468ec060ec603940b459abb5890767c7345d588659d5f2d099b69933f86
+size 1441920

ref/vjepa2-ac-vitg/frame0.pooled_mean.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e6eded56ec375a8531bc4ebe68ff26323161b967076bd25d36aa647d327a08a7
+size 5760

ref/vjepa2-ac-vitg/goalframe.context.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0cb6a6551f1c5667a319af53b6bd351c2350591cb31a59d04965e91460d17992
+size 1441920

ref/vjepa2-ac-vitg/goalframe.frames_u8.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b6368c136f846c95f85aa6833f9e7f088e2ee35d3ffddf8dcddaaaf3e1b61aec
+size 393344

ref/vjepa2-ac-vitg/goalframe.input.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:02d3d6383101ba5fd8968a17c3680242c1b4cff788c3d8bacd2d99c826159dfb
+size 1572992

ref/vjepa2-ac-vitg/goalframe.last_hidden_state.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9e08e6a93cc43972a1193ce912391cb67452f052dc2e71c40b68c9a50e89d19a
+size 1441920

ref/vjepa2-ac-vitg/goalframe.pooled_mean.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d5c50853f795fea020661547dade04487980c727a8346ea9687d3641e68f5d3d
+size 5760

ref/vjepa2-ac-vitg/manifest.json

@@ -0,0 +1,457 @@
+{
+ "model": "vjepa2-ac-vitg",
+ "source": "/home/overseer2/workdir/jepa.cpp/models/vjepa2_ac/vjepa2-ac-vitg.pt",
+ "dtype": "float32",
+ "npy_dtype_note": "all tensors float32 C-order except frames_u8 (uint8) and top5_idx (int64)",
+ "framework": {
+  "torch": "2.13.0+cpu",
+  "transformers": "5.16.1",
+  "python": "3.12.13",
+  "threads": 32,
+  "cpu": "x86_64"
+ },
+ "created": "2026-09-01T19:58:11+0300",
+ "source_url": "https://dl.fbaipublicfiles.com/vjepa2/vjepa2-ac-vitg.pt",
+ "code": "https://github.com/facebookresearch/vjepa2 (hubconf vjepa2_ac_vit_giant, src/models/ac_predictor.py, notebooks/utils/mpc_utils.py + world_model_wrapper.py), local clone at /home/overseer2/workdir/jepa.cpp/tmp/vjepa2-src",
+ "checkpoint": {
+  "top_level_keys": [
+   "encoder",
+   "predictor",
+   "opt",
+   "scaler",
+   "target_encoder",
+   "epoch",
+   "loss",
+   "batch_size",
+   "world_size",
+   "lr"
+  ],
+  "encoder_key_used": "encoder",
+  "key_cleanup": "strip 'module.' and 'backbone.'",
+  "epoch": 315,
+  "encoder_equals_target_encoder": true
+ },
+ "trajectory": {
+  "file": "notebooks/franka_example_traj.npz",
+  "frames": 2,
+  "frame_size_hw": [
+   256,
+   256
+  ],
+  "states": [
+   [
+    0.5804390907287598,
+    -0.0020212342496961355,
+    0.24755726754665375,
+    -3.066615581512451,
+    0.030529797077178955,
+    -1.9127511978149414,
+    0.9969713687896729
+   ],
+   [
+    0.6728084683418274,
+    0.028965970501303673,
+    0.33183684945106506,
+    -3.0514373779296875,
+    0.017709508538246155,
+    -1.9124950170516968,
+    0.9969713687896729
+   ]
+  ]
+ },
+ "hparams": {
+  "embed_dim": 1408,
+  "n_layer": 40,
+  "n_head": 22,
+  "ffn_dim": 6144,
+  "patch_size": 16,
+  "tubelet_size": 2,
+  "img_size": 256,
+  "n_frames": 64,
+  "ln_eps": 1e-06,
+  "act": "gelu_erf",
+  "cls_token": false,
+  "pos_type": "rope3d",
+  "rope_theta": 10000.0,
+  "tokens_per_frame": 256,
+  "pred": {
+   "kind": "ac",
+   "embed_dim": 1024,
+   "n_layer": 24,
+   "n_head": 16,
+   "head_dim": 64,
+   "ffn_dim": 4096,
+   "ln_eps": 1e-06,
+   "action_dim": 7,
+   "state_dim": 7,
+   "out_dim": 1408,
+   "grid_size": 16,
+   "n_cond_tokens": 2,
+   "cond_order": "action,state",
+   "frame_causal": true,
+   "use_extrinsics": false,
+   "attn_mask_shape": [
+    8256,
+    8256
+   ],
+   "rope_dims_per_axis": [
+    20,
+    20,
+    20
+   ],
+   "rope_grid_size": 16,
+   "rope_freq_layout": "tiled"
+  }
+ },
+ "preprocessing": {
+  "description": "app/vjepa_droid/transforms.py::make_transforms(random_horizontal_flip=False, random_resize_scale=(1,1), random_resize_aspect_ratio=(1,1), crop_size=256): random_resized_crop degenerates to a centre crop of the largest square followed by torch F.interpolate(bilinear, align_corners=False, no antialias) to 256x256, then (x - 255*mean) / (255*std) on the CTHW float tensor. On the 256x256 Franka renders both the crop and the resize are the identity.",
+  "resize": {
+   "shortest_edge": 256,
+   "resample": "bilinear",
+   "antialias": false,
+   "on_dtype": "float32",
+   "reference_order": "centre-crop to the largest square, THEN resize to 256"
+  },
+  "center_crop": 256,
+  "rescale": 0.00392156862745098,
+  "mean": [
+   0.485,
+   0.456,
+   0.406
+  ],
+  "std": [
+   0.229,
+   0.224,
+   0.225
+  ]
+ },
+ "world_model": {
+  "encode": "one observation frame -> a 2-frame clip (the frame repeated along T, tubelet 2) -> 256 tokens after the encoder's final LayerNorm; then a NON-AFFINE LayerNorm over the feature dim (torch F.layer_norm default eps 1e-5). world_model_wrapper.py:42-52.",
+  "predict": "predictor(context, actions, states) returns T*tpf rows; the world model keeps the LAST tpf (the next frame) and normalises them again before feeding them back. world_model_wrapper.py:56-64.",
+  "state": "the pose of frame t; the pose after a step is compute_new_pose(pose, action) = (xyz + d_xyz, euler_xyz(R(d) @ R(pose)), clip(gripper + d_gripper, 0, 1)). mpc_utils.py:166-190.",
+  "energy": "l1(a, b) = mean |a - b| over the flattened [tpf * D] final state and goal. mpc_utils.py:17-18 and the loss_fn of energy_landscape_example.ipynb.",
+  "candidates": "4 fixed action sequences of 2 steps: the trajectory's ground-truth first action poses_to_diff(states[0], states[1]) and that action shifted by +0.05 along x, y and z, each repeated over the horizon."
+ },
+ "outputs": {
+  "frames_u8": "the uint8 observation frame, repeated to the 2-frame clip, THWC",
+  "input": "NCTHW float32 (the 2-frame clip of one observation frame)",
+  "last_hidden_state": "encoder(input)[0]: [256, 1408] tokens after the final LN, h-major then w",
+  "context": "F.layer_norm(last_hidden_state) -- the predictor's input",
+  "goal": "context of the LAST trajectory frame",
+  "action/state": "[7] the action driving the step and the pose at the frame",
+  "pred_next": "predictor(context, action, state)[-256:] -- the next frame's latents, un-normalised",
+  "pred_seq": "predictor over T=2 frames, ALL T*tpf rows (row block t predicts frame t+1)",
+  "rollout": "[K, H, 256, 1408] normalised latents per candidate per step",
+  "rollout_energy": "[K] l1(rollout[:, -1], goal)",
+  "actions/states_seq": "[K, H, 7] the candidate actions and the poses compute_new_pose produced",
+  "next_state_ref": "[K, 7] compute_new_pose(state0, actions[:, 0]) -- the jepa_ac_next_state check"
+ },
+ "timing_s": {
+  "model_load": 8.727,
+  "forward_total": 3.22,
+  "forward_mean": 0.644,
+  "wall_total": 3.257
+ },
+ "samples": [
+  {
+   "name": "frame0",
+   "media": "franka_example_traj.npz[0]",
+   "timing_s": {
+    "preprocess_s": 0.0,
+    "forward_s": 0.5169
+   },
+   "tensors": {
+    "frames_u8": {
+     "file": "frame0.frames_u8.npy",
+     "shape": [
+      2,
+      256,
+      256,
+      3
+     ],
+     "dtype": "uint8",
+     "layout": "THWC uint8 (the observation frame, repeated to the 2-frame clip)"
+    },
+    "input": {
+     "file": "frame0.input.npy",
+     "shape": [
+      1,
+      3,
+      2,
+      256,
+      256
+     ],
+     "dtype": "float32",
+     "layout": "NCTHW (T=2, the frame repeated)"
+    },
+    "last_hidden_state": {
+     "file": "frame0.last_hidden_state.npy",
+     "shape": [
+      256,
+      1408
+     ],
+     "dtype": "float32",
+     "layout": "[256, D] h-major then w"
+    },
+    "pooled_mean": {
+     "file": "frame0.pooled_mean.npy",
+     "shape": [
+      1408
+     ],
+     "dtype": "float32",
+     "layout": "[D]"
+    },
+    "context": {
+     "file": "frame0.context.npy",
+     "shape": [
+      256,
+      1408
+     ],
+     "dtype": "float32",
+     "layout": "[256, D] after the non-affine LayerNorm"
+    }
+   },
+   "frame_index": 0,
+   "frames": 2
+  },
+  {
+   "name": "goalframe",
+   "media": "franka_example_traj.npz[1]",
+   "timing_s": {
+    "preprocess_s": 0.0,
+    "forward_s": 0.5064
+   },
+   "tensors": {
+    "frames_u8": {
+     "file": "goalframe.frames_u8.npy",
+     "shape": [
+      2,
+      256,
+      256,
+      3
+     ],
+     "dtype": "uint8",
+     "layout": "THWC uint8 (the observation frame, repeated to the 2-frame clip)"
+    },
+    "input": {
+     "file": "goalframe.input.npy",
+     "shape": [
+      1,
+      3,
+      2,
+      256,
+      256
+     ],
+     "dtype": "float32",
+     "layout": "NCTHW (T=2, the frame repeated)"
+    },
+    "last_hidden_state": {
+     "file": "goalframe.last_hidden_state.npy",
+     "shape": [
+      256,
+      1408
+     ],
+     "dtype": "float32",
+     "layout": "[256, D] h-major then w"
+    },
+    "pooled_mean": {
+     "file": "goalframe.pooled_mean.npy",
+     "shape": [
+      1408
+     ],
+     "dtype": "float32",
+     "layout": "[D]"
+    },
+    "context": {
+     "file": "goalframe.context.npy",
+     "shape": [
+      256,
+      1408
+     ],
+     "dtype": "float32",
+     "layout": "[256, D] after the non-affine LayerNorm"
+    }
+   },
+   "frame_index": 1,
+   "frames": 2
+  },
+  {
+   "name": "step1",
+   "media": "franka_example_traj.npz[0]",
+   "timing_s": {
+    "preprocess_s": 0.0,
+    "forward_s": 0.2069
+   },
+   "tensors": {
+    "context": {
+     "file": "step1.context.npy",
+     "shape": [
+      256,
+      1408
+     ],
+     "dtype": "float32",
+     "layout": "[256, D]"
+    },
+    "action": {
+     "file": "step1.action.npy",
+     "shape": [
+      7
+     ],
+     "dtype": "float32",
+     "layout": "[7]"
+    },
+    "state": {
+     "file": "step1.state.npy",
+     "shape": [
+      7
+     ],
+     "dtype": "float32",
+     "layout": "[7]"
+    },
+    "pred_next": {
+     "file": "step1.pred_next.npy",
+     "shape": [
+      256,
+      1408
+     ],
+     "dtype": "float32",
+     "layout": "[256, D] predictor output, T=1"
+    }
+   },
+   "frames": 1
+  },
+  {
+   "name": "step2",
+   "media": "franka_example_traj.npz[0]",
+   "timing_s": {
+    "preprocess_s": 0.0,
+    "forward_s": 0.3888
+   },
+   "tensors": {
+    "context_seq": {
+     "file": "step2.context_seq.npy",
+     "shape": [
+      512,
+      1408
+     ],
+     "dtype": "float32",
+     "layout": "[2*256, D]"
+    },
+    "action_seq": {
+     "file": "step2.action_seq.npy",
+     "shape": [
+      2,
+      7
+     ],
+     "dtype": "float32",
+     "layout": "[2, 7]"
+    },
+    "state_seq": {
+     "file": "step2.state_seq.npy",
+     "shape": [
+      2,
+      7
+     ],
+     "dtype": "float32",
+     "layout": "[2, 7]"
+    },
+    "pred_seq": {
+     "file": "step2.pred_seq.npy",
+     "shape": [
+      512,
+      1408
+     ],
+     "dtype": "float32",
+     "layout": "[2*256, D] all rows; block t predicts frame t+1"
+    }
+   },
+   "frames": 2
+  },
+  {
+   "name": "rollout",
+   "media": "franka_example_traj.npz",
+   "timing_s": {
+    "preprocess_s": 0.0,
+    "forward_s": 1.6008
+   },
+   "tensors": {
+    "context": {
+     "file": "rollout.context.npy",
+     "shape": [
+      256,
+      1408
+     ],
+     "dtype": "float32",
+     "layout": "[256, D] shared seed"
+    },
+    "goal": {
+     "file": "rollout.goal.npy",
+     "shape": [
+      256,
+      1408
+     ],
+     "dtype": "float32",
+     "layout": "[256, D]"
+    },
+    "state0": {
+     "file": "rollout.state0.npy",
+     "shape": [
+      7
+     ],
+     "dtype": "float32",
+     "layout": "[7]"
+    },
+    "actions": {
+     "file": "rollout.actions.npy",
+     "shape": [
+      4,
+      2,
+      7
+     ],
+     "dtype": "float32",
+     "layout": "[4, 2, 7]"
+    },
+    "states_seq": {
+     "file": "rollout.states_seq.npy",
+     "shape": [
+      4,
+      2,
+      7
+     ],
+     "dtype": "float32",
+     "layout": "[4, 2, 7]"
+    },
+    "rollout": {
+     "file": "rollout.rollout.npy",
+     "shape": [
+      4,
+      2,
+      256,
+      1408
+     ],
+     "dtype": "float32",
+     "layout": "[4, 2, 256, D]"
+    },
+    "rollout_energy": {
+     "file": "rollout.rollout_energy.npy",
+     "shape": [
+      4
+     ],
+     "dtype": "float32",
+     "layout": "[4]"
+    },
+    "next_state_ref": {
+     "file": "rollout.next_state_ref.npy",
+     "shape": [
+      4,
+      7
+     ],
+     "dtype": "float32",
+     "layout": "[4, 7]"
+    }
+   },
+   "candidates": 4,
+   "horizon": 2
+  }
+ ]
+}

ref/vjepa2-ac-vitg/rollout.actions.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cb9c6ed55eef6692d883a522842d6e1cb20ca306d57a765ffa4074859f87c595
+size 352

ref/vjepa2-ac-vitg/rollout.context.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6bddc62812f7ddb6447bb8da6b02b9e920e510ce495642f3719cf80da03af045
+size 1441920

ref/vjepa2-ac-vitg/rollout.goal.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0cb6a6551f1c5667a319af53b6bd351c2350591cb31a59d04965e91460d17992
+size 1441920

ref/vjepa2-ac-vitg/rollout.next_state_ref.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f37aa2d15cc0f0e95bffa12fadc03992c12fc461ad9481c8d2f339c58dad8ae5
+size 240

ref/vjepa2-ac-vitg/rollout.rollout.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9799a18580cc632b7193980358a4dcb3a435220422c185990211edba1806580b
+size 11534464

ref/vjepa2-ac-vitg/rollout.rollout_energy.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3159dd2a72816693cd28fc7f303f28856f7248faf285c140c9f1fdc549c70d38
+size 144

ref/vjepa2-ac-vitg/rollout.state0.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:22c6fc8e0d2fb5f2b4b57a887571fc1587df7b619f025035972340e84ce48f47
+size 156

ref/vjepa2-ac-vitg/rollout.states_seq.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8030997ff3d60c09272d40bab4ec24e29be8166baee6484c496e28de8c36b46b
+size 352

ref/vjepa2-ac-vitg/step1.action.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:faede2a2aea99785621f6ea07c021bf8d040423a994df95365ae9e0036d680ec
+size 156

ref/vjepa2-ac-vitg/step1.context.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6bddc62812f7ddb6447bb8da6b02b9e920e510ce495642f3719cf80da03af045
+size 1441920

ref/vjepa2-ac-vitg/step1.pred_next.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cb50055c423bcd93c54be66300a6faf9594d4a832334fc23bc8cb54bb2581b2f
+size 1441920

ref/vjepa2-ac-vitg/step1.state.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:22c6fc8e0d2fb5f2b4b57a887571fc1587df7b619f025035972340e84ce48f47
+size 156

ref/vjepa2-ac-vitg/step2.action_seq.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ae49af236d1641091a7e54bf53df72eb2ff33dc522d3564b819aeb38c94f453e
+size 184

ref/vjepa2-ac-vitg/step2.context_seq.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dbec8f7dd16f834c7e049ec26d96edea3b011bf6aec1731d112ca24e224a82e2
+size 2883712

ref/vjepa2-ac-vitg/step2.pred_seq.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:07ec6f44134122df7d8d096bd74ea19ab3f6cb6780bdc9ae995118bc3816554d
+size 2883712

ref/vjepa2-ac-vitg/step2.state_seq.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f85a9c66e4075d3e23ee0708a25292404acb47b12fd313fe9f7d6e9647c79d45
+size 184

ref/vjepa2-vitg-fpc64-256/archery_f16.frames_u8.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7bd1316d4cdf669019228e02cb9974c6e25188bd6da48150663aeda78b4b46f9
+size 8294528

ref/vjepa2-vitg-fpc64-256/archery_f16.input.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3addf20c07a53fad550fb96599f75da1f242121463eee4fe7d196765c8c3224d
+size 12583040

ref/vjepa2-vitg-fpc64-256/archery_f16.last_hidden_state.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:81a8019bd93c764571959f68f2e4519a0e9487b0d21b3012cc46719036cb5953
+size 11534464

ref/vjepa2-vitg-fpc64-256/archery_f16.pooled_mean.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fe29944df2e7416851b6701827a36b148d8024e27ce3045c99ed2f3d57f73f89
+size 5760

ref/vjepa2-vitg-fpc64-256/archery_f16.predictor_last_hidden_state.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c4134a8e0da1962415281bc64de3cb4fdebf8d94d9d7a42ba00e9980bfe2fee0
+size 11534464

ref/vjepa2-vitg-fpc64-256/archery_f64.frames_u8.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8364b537d5ffa132ec5a4b7311b42c8998e4fc08a9b6b01c229455708d2d670e
+size 33177728

ref/vjepa2-vitg-fpc64-256/archery_f64.input.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:119a2125d3082ca0fbca0d6d8cdc643b21d03502a00325b877a7873c2330f8a9
+size 50331776

ref/vjepa2-vitg-fpc64-256/archery_f64.last_hidden_state.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3c4c5e9563354c66e6d5859667066722b298f1e31d08d51bcb10a4cfd8fac20d
+size 46137472

ref/vjepa2-vitg-fpc64-256/archery_f64.pooled_mean.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fe8759557b474aac0c7120f47b36434fe7ef78cf154615cc14c7c0258dca1cd7
+size 5760

ref/vjepa2-vitg-fpc64-256/archery_f64.predictor_last_hidden_state.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d1559a4b7f3d9bbd6db6b4ad9d6645fe64d58891756710276df45d58be4b6e9e
+size 46137472

ref/vjepa2-vitg-fpc64-256/bowling_f16.frames_u8.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:157e4e592a57c89db7b13cb45c3993a8cdfe21bc71011f885302d679fd78ec20
+size 6220928

ref/vjepa2-vitg-fpc64-256/bowling_f16.input.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ac8908e2fcdb752524fb757ca3142a1cb17e97a5e5aac33262a8b1f1c8cba817
+size 12583040

ref/vjepa2-vitg-fpc64-256/bowling_f16.last_hidden_state.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2d9ae453cad05d78da0707f284bc7abd1fa03dc1e6633bc5785c8fcfc17e17a7
+size 11534464

ref/vjepa2-vitg-fpc64-256/bowling_f16.pooled_mean.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1e61e5e41f1dc35bb1fc9a5c595eff75f5d7cd80c190a8418d38c74ece1e007c
+size 5760

ref/vjepa2-vitg-fpc64-256/bowling_f16.predictor_last_hidden_state.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3fd043a7c298ccb235d446330762e90da26646c1ef2eb18f73a2f8d184be683d
+size 11534464

ref/vjepa2-vitg-fpc64-256/bowling_f64.frames_u8.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d5d9fc790de178d4318f0d8bd722b73d99f736771dc95f71bdd18f0393630ac2
+size 24883328

ref/vjepa2-vitg-fpc64-256/bowling_f64.input.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6c92a82ba4d036ffd403404d0349e7865804e261b5207c0bca379a42e964ab06
+size 50331776

ref/vjepa2-vitg-fpc64-256/bowling_f64.last_hidden_state.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:845f3703f00bda01925ea97ed766fd23534a115138585b4073f54a32b73f1257
+size 46137472

ref/vjepa2-vitg-fpc64-256/bowling_f64.pooled_mean.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ed16530465524f0768873da7a65688835bc00d64d1f11fbc54f2c0e27089adcf
+size 5760

ref/vjepa2-vitg-fpc64-256/bowling_f64.predictor_last_hidden_state.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1fe3548bffcabf6dcc6b0e691b274262265a8913389dbfdbcd5a7f8ae6a9bc78
+size 46137472

ref/vjepa2-vitg-fpc64-256/manifest.json

@@ -0,0 +1,545 @@
+{
+ "model": "vjepa2-vitg-fpc64-256",
+ "source": "/home/overseer2/workdir/jepa.cpp/models/facebook/vjepa2-vitg-fpc64-256",
+ "dtype": "float32",
+ "npy_dtype_note": "all tensors float32 C-order except frames_u8 (uint8) and top5_idx (int64)",
+ "framework": {
+  "torch": "2.13.0+cpu",
+  "transformers": "5.16.1",
+  "python": "3.12.13",
+  "threads": 32,
+  "cpu": "x86_64"
+ },
+ "created": "2026-09-01T19:10:46+0300",
+ "hf_id": "facebook/vjepa2-vitg-fpc64-256",
+ "hparams": {
+  "embed_dim": 1408,
+  "n_layer": 40,
+  "n_head": 22,
+  "ffn_dim": 6144,
+  "patch_size": 16,
+  "tubelet_size": 2,
+  "img_size": 256,
+  "n_frames": 64,
+  "ln_eps": 1e-06,
+  "act": "gelu_erf",
+  "qkv_bias": true,
+  "cls_token": false,
+  "pos_type": "rope3d",
+  "rope_theta": 10000.0,
+  "pred": {
+   "embed_dim": 384,
+   "n_layer": 12,
+   "n_head": 12,
+   "ffn_dim": 1536,
+   "n_mask_tokens": 10
+  }
+ },
+ "preprocessing": {
+  "description": "VJEPA2VideoProcessor: uint8 RGB frames (T,H,W,3) -> per-frame resize so the SHORT side = 292 (= int(crop*256/224)), aspect kept, torchvision.transforms.v2.functional.resize on the uint8 CHW tensor with BILINEAR + antialias=True (result rounded back to uint8) -> center crop 256x256 (top = int((H-256)/2), left = int((W-256)/2)) -> x/255 -> ImageNet mean/std -> pixel_values_videos [B, T, 3, H, W]",
+  "resize": {
+   "shortest_edge": 292,
+   "resample": "bilinear",
+   "antialias": true,
+   "on_dtype": "uint8"
+  },
+  "center_crop": 256,
+  "rescale": 0.00392156862745098,
+  "mean": [
+   0.485,
+   0.456,
+   0.406
+  ],
+  "std": [
+   0.229,
+   0.224,
+   0.225
+  ],
+  "frame_sampling": "idx = round(linspace(0, T_total-1, n)) over all decoded frames (PyAV rgb24); indices stored per sample",
+  "resampling_note": "resampling = torchvision.transforms.v2.functional.resize on the uint8 CHW tensor with antialias=True (result rounded back to uint8), exactly what transformers>=5 image/video processors (TorchvisionBackend) do; verified to 2.4e-7 against a re-implementation, whereas PIL resampling differs by up to 1.8e-2 (1-2 uint8 levels)",
+  "processor": {
+   "class": "VJEPA2VideoProcessor",
+   "backend": "torchvision",
+   "config": {
+    "crop_size": {
+     "height": 256,
+     "width": 256
+    },
+    "data_format": "channels_first",
+    "default_to_square": true,
+    "do_center_crop": true,
+    "do_normalize": true,
+    "do_rescale": true,
+    "do_resize": true,
+    "image_mean": [
+     0.485,
+     0.456,
+     0.406
+    ],
+    "image_std": [
+     0.229,
+     0.224,
+     0.225
+    ],
+    "resample": 2,
+    "rescale_factor": 0.00392156862745098,
+    "size": {
+     "shortest_edge": 292
+    },
+    "video_processor_type": "VJEPA2VideoProcessor",
+    "return_metadata": false
+   }
+  }
+ },
+ "outputs": {
+  "frames_u8": "the sampled RGB frames fed to the processor, THWC uint8",
+  "input": "processor pixel_values_videos, layout NTCHW (batch, frames, channels, H, W); the model permutes to NCTHW internally",
+  "last_hidden_state": "VJEPA2Model last_hidden_state[0]: [N, 1408] with N = T/2 * 16 * 16 tokens, order t-major then h then w, after final LN",
+  "pooled_mean": "mean over tokens",
+  "predictor_last_hidden_state": "VJEPA2Model(...).predictor_output.last_hidden_state[0] with the DEFAULT masks (context_mask = target_mask = arange(N), i.e. full context, predict all tokens): [N, 1408] = predictor_proj(predictor_norm(...)) for the N target (mask) tokens"
+ },
+ "timing_s": {
+  "model_load": 0.241,
+  "forward_total": 42.5,
+  "forward_mean": 10.625,
+  "wall_total": 43.308
+ },
+ "samples": [
+  {
+   "name": "archery_f64",
+   "media": "archery.mp4",
+   "timing_s": {
+    "decode_s": 0.4155,
+    "preprocess_s": 0.0455,
+    "forward_s": 18.8015
+   },
+   "tensors": {
+    "frames_u8": {
+     "file": "archery_f64.frames_u8.npy",
+     "shape": [
+      64,
+      360,
+      480,
+      3
+     ],
+     "dtype": "uint8",
+     "layout": "THWC uint8"
+    },
+    "input": {
+     "file": "archery_f64.input.npy",
+     "shape": [
+      1,
+      64,
+      3,
+      256,
+      256
+     ],
+     "dtype": "float32",
+     "layout": "NTCHW"
+    },
+    "last_hidden_state": {
+     "file": "archery_f64.last_hidden_state.npy",
+     "shape": [
+      8192,
+      1408
+     ],
+     "dtype": "float32",
+     "layout": "[N_tokens, D] t-major,h,w"
+    },
+    "pooled_mean": {
+     "file": "archery_f64.pooled_mean.npy",
+     "shape": [
+      1408
+     ],
+     "dtype": "float32",
+     "layout": "[D]"
+    },
+    "predictor_last_hidden_state": {
+     "file": "archery_f64.predictor_last_hidden_state.npy",
+     "shape": [
+      8192,
+      1408
+     ],
+     "dtype": "float32",
+     "layout": "[N_tokens, D]"
+    }
+   },
+   "frames": 64,
+   "frame_indices": [
+    0,
+    5,
+    9,
+    14,
+    19,
+    24,
+    28,
+    33,
+    38,
+    43,
+    47,
+    52,
+    57,
+    62,
+    66,
+    71,
+    76,
+    81,
+    85,
+    90,
+    95,
+    100,
+    104,
+    109,
+    114,
+    119,
+    123,
+    128,
+    133,
+    138,
+    142,
+    147,
+    152,
+    157,
+    161,
+    166,
+    171,
+    176,
+    180,
+    185,
+    190,
+    195,
+    199,
+    204,
+    209,
+    214,
+    218,
+    223,
+    228,
+    233,
+    237,
+    242,
+    247,
+    252,
+    256,
+    261,
+    266,
+    271,
+    275,
+    280,
+    285,
+    290,
+    294,
+    299
+   ],
+   "n_frames_total": 300,
+   "fps": 29.97002997002997,
+   "frame_size_hw": [
+    360,
+    480
+   ]
+  },
+  {
+   "name": "archery_f16",
+   "media": "archery.mp4",
+   "timing_s": {
+    "decode_s": 0.4155,
+    "preprocess_s": 0.009,
+    "forward_s": 2.7045
+   },
+   "tensors": {
+    "frames_u8": {
+     "file": "archery_f16.frames_u8.npy",
+     "shape": [
+      16,
+      360,
+      480,
+      3
+     ],
+     "dtype": "uint8",
+     "layout": "THWC uint8"
+    },
+    "input": {
+     "file": "archery_f16.input.npy",
+     "shape": [
+      1,
+      16,
+      3,
+      256,
+      256
+     ],
+     "dtype": "float32",
+     "layout": "NTCHW"
+    },
+    "last_hidden_state": {
+     "file": "archery_f16.last_hidden_state.npy",
+     "shape": [
+      2048,
+      1408
+     ],
+     "dtype": "float32",
+     "layout": "[N_tokens, D] t-major,h,w"
+    },
+    "pooled_mean": {
+     "file": "archery_f16.pooled_mean.npy",
+     "shape": [
+      1408
+     ],
+     "dtype": "float32",
+     "layout": "[D]"
+    },
+    "predictor_last_hidden_state": {
+     "file": "archery_f16.predictor_last_hidden_state.npy",
+     "shape": [
+      2048,
+      1408
+     ],
+     "dtype": "float32",
+     "layout": "[N_tokens, D]"
+    }
+   },
+   "frames": 16,
+   "frame_indices": [
+    0,
+    20,
+    40,
+    60,
+    80,
+    100,
+    120,
+    140,
+    159,
+    179,
+    199,
+    219,
+    239,
+    259,
+    279,
+    299
+   ],
+   "n_frames_total": 300,
+   "fps": 29.97002997002997,
+   "frame_size_hw": [
+    360,
+    480
+   ]
+  },
+  {
+   "name": "bowling_f64",
+   "media": "bowling.mp4",
+   "timing_s": {
+    "decode_s": 0.1721,
+    "preprocess_s": 0.0527,
+    "forward_s": 18.4933
+   },
+   "tensors": {
+    "frames_u8": {
+     "file": "bowling_f64.frames_u8.npy",
+     "shape": [
+      64,
+      270,
+      480,
+      3
+     ],
+     "dtype": "uint8",
+     "layout": "THWC uint8"
+    },
+    "input": {
+     "file": "bowling_f64.input.npy",
+     "shape": [
+      1,
+      64,
+      3,
+      256,
+      256
+     ],
+     "dtype": "float32",
+     "layout": "NTCHW"
+    },
+    "last_hidden_state": {
+     "file": "bowling_f64.last_hidden_state.npy",
+     "shape": [
+      8192,
+      1408
+     ],
+     "dtype": "float32",
+     "layout": "[N_tokens, D] t-major,h,w"
+    },
+    "pooled_mean": {
+     "file": "bowling_f64.pooled_mean.npy",
+     "shape": [
+      1408
+     ],
+     "dtype": "float32",
+     "layout": "[D]"
+    },
+    "predictor_last_hidden_state": {
+     "file": "bowling_f64.predictor_last_hidden_state.npy",
+     "shape": [
+      8192,
+      1408
+     ],
+     "dtype": "float32",
+     "layout": "[N_tokens, D]"
+    }
+   },
+   "frames": 64,
+   "frame_indices": [
+    0,
+    2,
+    5,
+    7,
+    9,
+    12,
+    14,
+    17,
+    19,
+    21,
+    24,
+    26,
+    28,
+    31,
+    33,
+    35,
+    38,
+    40,
+    43,
+    45,
+    47,
+    50,
+    52,
+    54,
+    57,
+    59,
+    61,
+    64,
+    66,
+    69,
+    71,
+    73,
+    76,
+    78,
+    80,
+    83,
+    85,
+    88,
+    90,
+    92,
+    95,
+    97,
+    99,
+    102,
+    104,
+    106,
+    109,
+    111,
+    114,
+    116,
+    118,
+    121,
+    123,
+    125,
+    128,
+    130,
+    132,
+    135,
+    137,
+    140,
+    142,
+    144,
+    147,
+    149
+   ],
+   "n_frames_total": 150,
+   "fps": 15.0,
+   "frame_size_hw": [
+    270,
+    480
+   ]
+  },
+  {
+   "name": "bowling_f16",
+   "media": "bowling.mp4",
+   "timing_s": {
+    "decode_s": 0.1721,
+    "preprocess_s": 0.0109,
+    "forward_s": 2.5007
+   },
+   "tensors": {
+    "frames_u8": {
+     "file": "bowling_f16.frames_u8.npy",
+     "shape": [
+      16,
+      270,
+      480,
+      3
+     ],
+     "dtype": "uint8",
+     "layout": "THWC uint8"
+    },
+    "input": {
+     "file": "bowling_f16.input.npy",
+     "shape": [
+      1,
+      16,
+      3,
+      256,
+      256
+     ],
+     "dtype": "float32",
+     "layout": "NTCHW"
+    },
+    "last_hidden_state": {
+     "file": "bowling_f16.last_hidden_state.npy",
+     "shape": [
+      2048,
+      1408
+     ],
+     "dtype": "float32",
+     "layout": "[N_tokens, D] t-major,h,w"
+    },
+    "pooled_mean": {
+     "file": "bowling_f16.pooled_mean.npy",
+     "shape": [
+      1408
+     ],
+     "dtype": "float32",
+     "layout": "[D]"
+    },
+    "predictor_last_hidden_state": {
+     "file": "bowling_f16.predictor_last_hidden_state.npy",
+     "shape": [
+      2048,
+      1408
+     ],
+     "dtype": "float32",
+     "layout": "[N_tokens, D]"
+    }
+   },
+   "frames": 16,
+   "frame_indices": [
+    0,
+    10,
+    20,
+    30,
+    40,
+    50,
+    60,
+    70,
+    79,
+    89,
+    99,
+    109,
+    119,
+    129,
+    139,
+    149
+   ],
+   "n_frames_total": 150,
+   "fps": 15.0,
+   "frame_size_hw": [
+    270,
+    480
+   ]
+  }
+ ]
+}