Upload folder using huggingface_hub

export_weights.py

@@ -0,0 +1,164 @@
+#!/usr/bin/env python3
+"""Export the six named LightVAE components without changing source files."""
+
+import argparse
+from collections import Counter
+from datetime import datetime, timezone
+import hashlib
+import json
+import os
+from pathlib import Path
+import shutil
+import tempfile
+
+import torch
+from safetensors import safe_open
+from safetensors.torch import save_file
+
+
+COMPONENTS = (
+    ("wan21_full_depth_latest.pt", "lightvae-pro-wan21-decoder"),
+    ("wan21_compact_latest.pt", "lightvae-lite-wan21-decoder"),
+    ("wan22_full_depth_latest.pt", "lightvae-pro-wan22-decoder"),
+    ("wan22_compact_latest.pt", "lightvae-lite-wan22-decoder"),
+    ("minimax_h3_lightvae_decoder.safetensors", "lightvae-lite-h3-decoder"),
+    ("minimax_h3_lightvae_encoder.safetensors", "lightvae-lite-h3-encoder"),
+)
+
+
+def sha256(path):
+    digest = hashlib.sha256()
+    with path.open("rb") as stream:
+        for chunk in iter(lambda: stream.read(8 * 1024 * 1024), b""):
+            digest.update(chunk)
+    return digest.hexdigest()
+
+
+def write_json(path, value):
+    with path.open("x", encoding="utf-8") as stream:
+        json.dump(value, stream, ensure_ascii=False, indent=2, allow_nan=False)
+        stream.write("\n")
+
+
+def inspect_safetensors(path):
+    with safe_open(path, framework="pt", device="cpu") as saved:
+        keys = sorted(saved.keys())
+        dtypes = Counter(saved.get_slice(key).get_dtype() for key in keys)
+    return {
+        "tensor_count": len(keys),
+        "tensor_prefix_counts": dict(Counter(key.split(".")[0] for key in keys)),
+        "tensor_dtype_counts": dict(dtypes),
+    }
+
+
+def convert_wan(source, target):
+    saved = torch.load(source, weights_only=True, map_location="cpu", mmap=True)
+    state = saved["model"]
+    if not isinstance(state, dict) or not state:
+        raise ValueError(f"Missing model state: {source}")
+    if any(not isinstance(key, str) or not isinstance(value, torch.Tensor)
+           for key, value in state.items()):
+        raise TypeError(f"Non-tensor model entry: {source}")
+    checkpoint_metadata = {key: value for key, value in saved.items() if key != "model"}
+    encoded = json.dumps(checkpoint_metadata, ensure_ascii=False, allow_nan=False)
+    tensors = {key: value.detach().cpu().contiguous() for key, value in state.items()}
+    save_file(tensors, target, metadata={
+        "format": "pt",
+        "model_type": "lightvae_decoder",
+        "checkpoint_metadata": encoded,
+    })
+    with safe_open(target, framework="pt", device="cpu") as restored:
+        if set(restored.keys()) != set(state):
+            raise AssertionError(f"Tensor keys changed: {source}")
+        if json.loads(restored.metadata()["checkpoint_metadata"]) != json.loads(encoded):
+            raise AssertionError(f"Checkpoint metadata changed: {source}")
+        for key, expected in tensors.items():
+            actual = restored.get_tensor(key)
+            if actual.dtype != expected.dtype or actual.shape != expected.shape:
+                raise AssertionError(f"Shape/dtype changed: {source}:{key}")
+            # Compare bytes, including NaN payloads and signed zeros.
+            if not torch.equal(actual.reshape(-1).view(torch.uint8),
+                               expected.reshape(-1).view(torch.uint8)):
+                raise AssertionError(f"Tensor values changed: {source}:{key}")
+    return checkpoint_metadata
+
+
+def main():
+    parser = argparse.ArgumentParser(description=__doc__)
+    parser.add_argument("--source-dir", type=Path, default=Path(
+        "/data/nvme6/gushiqiao/codes/latest/vae_prune/models"))
+    parser.add_argument("--output-dir", type=Path, default=Path(__file__).resolve().parent)
+    args = parser.parse_args()
+    torch.set_num_threads(4)
+    destination = args.output_dir.resolve()
+    destination.mkdir(parents=True, exist_ok=True)
+    output_names = ["manifest.json"]
+    for source_name, stem in COMPONENTS:
+        if not (args.source_dir / source_name).is_file():
+            raise FileNotFoundError(args.source_dir / source_name)
+        output_names.extend((stem + ".safetensors", stem + ".metadata.json"))
+    for name in output_names:
+        if (destination / name).exists() or (destination / name).is_symlink():
+            raise FileExistsError(f"Refusing to overwrite: {destination / name}")
+
+    manifest = {
+        "format_version": 1,
+        "created_utc": datetime.now(timezone.utc).isoformat(),
+        "source_files_unchanged": True,
+        "precision_policy": "Preserve all tensor names, shapes, dtypes and values.",
+        "loading_notes": {
+            "wan": "Use safetensors.torch.load_file for the model state. Original "
+                   "checkpoint fields (family, kept, packed_records, provenance, step, etc.) "
+                   "are preserved in the checkpoint_metadata header JSON and sidecar. "
+                   "The existing torch.load-based .pt loader cannot load these files unchanged.",
+            "h3": "Byte-for-byte copy of the original component checkpoint, including header metadata.",
+            "components": "Decoder files do not include an encoder. The H3 encoder is pruned, not original.",
+            "wan21_lite_source": "Uses wan21_compact_latest.pt as requested, not the later recovery/student.pt.",
+        },
+        "weights": [],
+    }
+    with tempfile.TemporaryDirectory(prefix=".export-", dir=destination) as work:
+        staging = Path(work)
+        for source_name, stem in COMPONENTS:
+            source = (args.source_dir / source_name).resolve()
+            target = staging / (stem + ".safetensors")
+            print(f"Exporting {source.name} -> {target.name}", flush=True)
+            initial_stat = source.stat()
+            source_digest = sha256(source)
+            if source.suffix == ".pt":
+                metadata = convert_wan(source, target)
+                validation = "all_tensor_keys_shapes_dtypes_and_bytes_equal; metadata_equal"
+            else:
+                shutil.copyfile(source, target)
+                with safe_open(source, framework="pt", device="cpu") as original:
+                    metadata = original.metadata()
+                validation = "whole_file_sha256_equal_including_header_metadata"
+            target_digest = sha256(target)
+            if source.suffix == ".safetensors" and source_digest != target_digest:
+                raise AssertionError(f"Copy verification failed: {source}")
+            final_stat = source.stat()
+            if (initial_stat.st_size, initial_stat.st_mtime_ns) != (
+                    final_stat.st_size, final_stat.st_mtime_ns):
+                raise RuntimeError(f"Source changed during export: {source}")
+            metadata_name = stem + ".metadata.json"
+            write_json(staging / metadata_name, metadata)
+            manifest["weights"].append({
+                "file": target.name,
+                "metadata_file": metadata_name,
+                "source": str(source),
+                "source_sha256": source_digest,
+                "sha256": target_digest,
+                "size_bytes": target.stat().st_size,
+                "validation": validation,
+                **inspect_safetensors(target),
+            })
+            print(f"Verified {target.name}", flush=True)
+        write_json(staging / "manifest.json", manifest)
+        # Hard-link promotion refuses overwriting even if another process raced us.
+        for name in output_names[1:] + ["manifest.json"]:
+            os.link(staging / name, destination / name)
+    print(f"Done: {len(COMPONENTS)} verified weights in {destination}", flush=True)
+
+
+if __name__ == "__main__":
+    main()

lightvae-lite-h3-decoder.metadata.json

@@ -0,0 +1,5 @@
+{
+  "model_type": "minimax_h3_pruned_vae",
+  "format": "pt",
+  "architecture": "{\"kept_layers\":[1,3,6,9,13,16,18,21,26,27,30,34],\"search\":null,\"teacher_config\":{\"_class_name\":\"AutoencoderKLMiniMaxH3\",\"_diffusers_version\":\"0.36.0.dev0\",\"block_out_channels\":[128,256,256,512,512,1024],\"clip_length\":17,\"decoder_attention_head_dim\":64,\"decoder_ffn_mult\":4,\"decoder_norm_eps\":1e-05,\"decoder_num_attention_heads\":32,\"decoder_num_layers\":36,\"decoder_num_register_tokens\":4,\"decoder_rope_dim_ratio\":0.75,\"decoder_rope_theta\":100.0,\"in_channels\":3,\"latent_channels\":24,\"latents_mean\":[0.858090341091156,-0.9606591463088989,1.0661640167236328,-0.5090325474739075,-0.2727581858634949,-1.3675414323806763,-0.2553254961967468,-0.26907554268836975,-0.5376840829849243,-0.0464097298681736,0.6657370328903198,0.19690127670764923,-0.5460608005523682,-0.4035342037677765,-0.23683024942874908,0.25928452610969543,-0.30133944749832153,0.211341992020607,-1.1206848621368408,0.3581933379173279,-0.04225143790245056,0.2604829967021942,0.22864092886447906,0.7056031823158264],\"latents_std\":[1.2223774194717407,1.2767263650894165,1.6831774711608887,1.7549455165863037,1.5636216402053833,2.194143533706665,0.9653137922286987,1.0569885969161987,0.841948926448822,0.7729952931404114,1.8955937623977661,0.946841835975647,0.7996809482574463,0.44988900423049927,0.7197399735450745,0.6936293244361877,2.961095094680786,2.7694199085235596,3.0496184825897217,2.1088054180145264,3.276226282119751,3.1627357006073,2.2816812992095947,2.6127843856811523],\"layers_per_block\":2,\"norm_eps\":1e-06,\"norm_num_groups\":32,\"out_channels\":3,\"spatial_downsample_factors\":[2,2,2,2,1,1],\"spatial_padding_mode\":\"reflect\",\"temporal_downsample_factors\":[1,2,2,1,1,1],\"token_drop\":3},\"tile_sample_min_height\":256,\"tile_sample_min_overlap_height\":64,\"tile_sample_min_overlap_width\":64,\"tile_sample_min_width\":256}"
+}

lightvae-lite-h3-decoder.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d060ad616a33f422fcb8395f525d6d71185b5ddb3e016b31cc210ba2c9c06ee9
+size 3248356840

lightvae-lite-h3-encoder.metadata.json

@@ -0,0 +1,5 @@
+{
+  "architecture": "{\"kept_residual_indices\":[0,2,4,6,8,11],\"search\":null,\"teacher_config\":{\"_class_name\":\"AutoencoderKLMiniMaxH3\",\"_diffusers_version\":\"0.36.0.dev0\",\"block_out_channels\":[128,256,256,512,512,1024],\"clip_length\":17,\"decoder_attention_head_dim\":64,\"decoder_ffn_mult\":4,\"decoder_norm_eps\":1e-05,\"decoder_num_attention_heads\":32,\"decoder_num_layers\":36,\"decoder_num_register_tokens\":4,\"decoder_rope_dim_ratio\":0.75,\"decoder_rope_theta\":100.0,\"in_channels\":3,\"latent_channels\":24,\"latents_mean\":[0.858090341091156,-0.9606591463088989,1.0661640167236328,-0.5090325474739075,-0.2727581858634949,-1.3675414323806763,-0.2553254961967468,-0.26907554268836975,-0.5376840829849243,-0.0464097298681736,0.6657370328903198,0.19690127670764923,-0.5460608005523682,-0.4035342037677765,-0.23683024942874908,0.25928452610969543,-0.30133944749832153,0.211341992020607,-1.1206848621368408,0.3581933379173279,-0.04225143790245056,0.2604829967021942,0.22864092886447906,0.7056031823158264],\"latents_std\":[1.2223774194717407,1.2767263650894165,1.6831774711608887,1.7549455165863037,1.5636216402053833,2.194143533706665,0.9653137922286987,1.0569885969161987,0.841948926448822,0.7729952931404114,1.8955937623977661,0.946841835975647,0.7996809482574463,0.44988900423049927,0.7197399735450745,0.6936293244361877,2.961095094680786,2.7694199085235596,3.0496184825897217,2.1088054180145264,3.276226282119751,3.1627357006073,2.2816812992095947,2.6127843856811523],\"layers_per_block\":2,\"norm_eps\":1e-06,\"norm_num_groups\":32,\"out_channels\":3,\"spatial_downsample_factors\":[2,2,2,2,1,1],\"spatial_padding_mode\":\"reflect\",\"temporal_downsample_factors\":[1,2,2,1,1,1],\"token_drop\":3},\"tile_sample_min_height\":256,\"tile_sample_min_overlap_height\":64,\"tile_sample_min_overlap_width\":64,\"tile_sample_min_width\":256,\"use_tiling\":true}",
+  "format": "pt",
+  "model_type": "minimax_h3_pruned_encoder"
+}

lightvae-lite-h3-encoder.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4b0e5af102da19324c64df40abbd95008d8acb0b9cf76e4704d74c8ac20cefac
+size 406335344

lightvae-lite-wan21-decoder.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05cbfff899d1bd8ce5afa0cc5ed03dcd01dd95876a76b1fb7df5427353f822ce
+size 70875748

lightvae-lite-wan22-decoder.metadata.json

@@ -0,0 +1,231 @@
+{
+  "format_version": 1,
+  "name": "wan22_compact",
+  "family": "wan22",
+  "kept": [
+    0,
+    4,
+    5,
+    8,
+    11
+  ],
+  "packed_records": [
+    {
+      "path": "conv1",
+      "shape": [
+        1024,
+        48,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 290,
+      "bias": true,
+      "pure2d": false,
+      "fraction": 0.6
+    },
+    {
+      "path": "middle.0.residual.2",
+      "shape": [
+        1024,
+        1024,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 420,
+      "bias": true,
+      "pure2d": false,
+      "fraction": 0.6
+    },
+    {
+      "path": "middle.0.residual.6",
+      "shape": [
+        1024,
+        1024,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 434,
+      "bias": true,
+      "pure2d": false,
+      "fraction": 0.6
+    },
+    {
+      "path": "upsamples.0.upsamples.2.residual.2",
+      "shape": [
+        1024,
+        1024,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 411,
+      "bias": true,
+      "pure2d": false,
+      "fraction": 0.6
+    },
+    {
+      "path": "upsamples.0.upsamples.2.residual.6",
+      "shape": [
+        1024,
+        1024,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 452,
+      "bias": true,
+      "pure2d": false,
+      "fraction": 0.6
+    },
+    {
+      "path": "upsamples.0.upsamples.3.time_conv",
+      "shape": [
+        2048,
+        1024,
+        3,
+        1,
+        1
+      ],
+      "padding": 0,
+      "rank": 533,
+      "bias": true,
+      "pure2d": false,
+      "fraction": 0.6
+    },
+    {
+      "path": "upsamples.1.upsamples.0.residual.2",
+      "shape": [
+        1024,
+        1024,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 448,
+      "bias": true,
+      "pure2d": false,
+      "fraction": 0.6
+    },
+    {
+      "path": "upsamples.1.upsamples.0.residual.6",
+      "shape": [
+        1024,
+        1024,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 456,
+      "bias": true,
+      "pure2d": false,
+      "fraction": 0.6
+    },
+    {
+      "path": "upsamples.1.upsamples.3.time_conv",
+      "shape": [
+        2048,
+        1024,
+        3,
+        1,
+        1
+      ],
+      "padding": 0,
+      "rank": 590,
+      "bias": true,
+      "pure2d": false,
+      "fraction": 0.6
+    },
+    {
+      "path": "upsamples.2.upsamples.0.residual.2",
+      "shape": [
+        512,
+        1024,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 0,
+      "bias": true,
+      "pure2d": true,
+      "fraction": 0.6
+    },
+    {
+      "path": "upsamples.2.upsamples.0.residual.6",
+      "shape": [
+        512,
+        512,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 0,
+      "bias": true,
+      "pure2d": true,
+      "fraction": 0.6
+    },
+    {
+      "path": "upsamples.3.upsamples.0.residual.2",
+      "shape": [
+        256,
+        512,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 0,
+      "bias": true,
+      "pure2d": true,
+      "fraction": 0.6
+    },
+    {
+      "path": "upsamples.3.upsamples.0.residual.6",
+      "shape": [
+        256,
+        256,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 0,
+      "bias": true,
+      "pure2d": true,
+      "fraction": 0.6
+    },
+    {
+      "path": "head.2",
+      "shape": [
+        12,
+        256,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 7,
+      "bias": true,
+      "pure2d": false,
+      "fraction": 0.6
+    }
+  ],
+  "provenance": {
+    "source": "/data/nvme3/gushiqiao/codes/wan22_11111_sum60_tail2d_all_teacher_gtl1_20260917/checkpoints/checkpoint-001000.pt",
+    "source_step": 1000,
+    "source_sha256": "3808b89d9d0c513f8290efb4d76b90ef756cf0c2667f080b45bdabff6371742b",
+    "source_bytes": 1904990651
+  },
+  "step": 3600
+}

lightvae-lite-wan22-decoder.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:455846ab42fe580b82209aec2bee84774860e29a6a66673ec6effe99400b6197
+size 609707592

lightvae-pro-wan21-decoder.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:15597278a84b3b416f5f41552ed38a6145a403e5792760979f7bb74f7fb5c0eb
+size 179047668

lightvae-pro-wan22-decoder.metadata.json

@@ -0,0 +1,510 @@
+{
+  "format_version": 1,
+  "name": "wan22_full_depth",
+  "family": "wan22",
+  "kept": [
+    0,
+    1,
+    2,
+    3,
+    4,
+    5,
+    6,
+    7,
+    8,
+    9,
+    10,
+    11,
+    12,
+    13
+  ],
+  "packed_records": [
+    {
+      "path": "conv1",
+      "shape": [
+        1024,
+        48,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 290,
+      "bias": true,
+      "pure2d": false,
+      "fraction": 0.6
+    },
+    {
+      "path": "middle.0.residual.2",
+      "shape": [
+        1024,
+        1024,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 420,
+      "bias": true,
+      "pure2d": false,
+      "fraction": 0.6
+    },
+    {
+      "path": "middle.0.residual.6",
+      "shape": [
+        1024,
+        1024,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 434,
+      "bias": true,
+      "pure2d": false,
+      "fraction": 0.6
+    },
+    {
+      "path": "middle.2.residual.2",
+      "shape": [
+        1024,
+        1024,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 433,
+      "bias": true,
+      "pure2d": false,
+      "fraction": 0.6
+    },
+    {
+      "path": "middle.2.residual.6",
+      "shape": [
+        1024,
+        1024,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 441,
+      "bias": true,
+      "pure2d": false,
+      "fraction": 0.6
+    },
+    {
+      "path": "upsamples.0.upsamples.0.residual.2",
+      "shape": [
+        1024,
+        1024,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 406,
+      "bias": true,
+      "pure2d": false,
+      "fraction": 0.6
+    },
+    {
+      "path": "upsamples.0.upsamples.0.residual.6",
+      "shape": [
+        1024,
+        1024,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 427,
+      "bias": true,
+      "pure2d": false,
+      "fraction": 0.6
+    },
+    {
+      "path": "upsamples.0.upsamples.1.residual.2",
+      "shape": [
+        1024,
+        1024,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 430,
+      "bias": true,
+      "pure2d": false,
+      "fraction": 0.6
+    },
+    {
+      "path": "upsamples.0.upsamples.1.residual.6",
+      "shape": [
+        1024,
+        1024,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 456,
+      "bias": true,
+      "pure2d": false,
+      "fraction": 0.6
+    },
+    {
+      "path": "upsamples.0.upsamples.2.residual.2",
+      "shape": [
+        1024,
+        1024,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 411,
+      "bias": true,
+      "pure2d": false,
+      "fraction": 0.6
+    },
+    {
+      "path": "upsamples.0.upsamples.2.residual.6",
+      "shape": [
+        1024,
+        1024,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 452,
+      "bias": true,
+      "pure2d": false,
+      "fraction": 0.6
+    },
+    {
+      "path": "upsamples.0.upsamples.3.time_conv",
+      "shape": [
+        2048,
+        1024,
+        3,
+        1,
+        1
+      ],
+      "padding": 0,
+      "rank": 533,
+      "bias": true,
+      "pure2d": false,
+      "fraction": 0.6
+    },
+    {
+      "path": "upsamples.1.upsamples.0.residual.2",
+      "shape": [
+        1024,
+        1024,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 448,
+      "bias": true,
+      "pure2d": false,
+      "fraction": 0.6
+    },
+    {
+      "path": "upsamples.1.upsamples.0.residual.6",
+      "shape": [
+        1024,
+        1024,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 457,
+      "bias": true,
+      "pure2d": false,
+      "fraction": 0.6
+    },
+    {
+      "path": "upsamples.1.upsamples.1.residual.2",
+      "shape": [
+        1024,
+        1024,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 475,
+      "bias": true,
+      "pure2d": false,
+      "fraction": 0.6
+    },
+    {
+      "path": "upsamples.1.upsamples.1.residual.6",
+      "shape": [
+        1024,
+        1024,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 485,
+      "bias": true,
+      "pure2d": false,
+      "fraction": 0.6
+    },
+    {
+      "path": "upsamples.1.upsamples.2.residual.2",
+      "shape": [
+        1024,
+        1024,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 467,
+      "bias": true,
+      "pure2d": false,
+      "fraction": 0.6
+    },
+    {
+      "path": "upsamples.1.upsamples.2.residual.6",
+      "shape": [
+        1024,
+        1024,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 477,
+      "bias": true,
+      "pure2d": false,
+      "fraction": 0.6
+    },
+    {
+      "path": "upsamples.1.upsamples.3.time_conv",
+      "shape": [
+        2048,
+        1024,
+        3,
+        1,
+        1
+      ],
+      "padding": 0,
+      "rank": 590,
+      "bias": true,
+      "pure2d": false,
+      "fraction": 0.6
+    },
+    {
+      "path": "upsamples.2.upsamples.0.residual.2",
+      "shape": [
+        512,
+        1024,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 0,
+      "bias": true,
+      "pure2d": true,
+      "fraction": 0.6
+    },
+    {
+      "path": "upsamples.2.upsamples.0.residual.6",
+      "shape": [
+        512,
+        512,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 0,
+      "bias": true,
+      "pure2d": true,
+      "fraction": 0.6
+    },
+    {
+      "path": "upsamples.2.upsamples.1.residual.2",
+      "shape": [
+        512,
+        512,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 0,
+      "bias": true,
+      "pure2d": true,
+      "fraction": 0.6
+    },
+    {
+      "path": "upsamples.2.upsamples.1.residual.6",
+      "shape": [
+        512,
+        512,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 0,
+      "bias": true,
+      "pure2d": true,
+      "fraction": 0.6
+    },
+    {
+      "path": "upsamples.2.upsamples.2.residual.2",
+      "shape": [
+        512,
+        512,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 0,
+      "bias": true,
+      "pure2d": true,
+      "fraction": 0.6
+    },
+    {
+      "path": "upsamples.2.upsamples.2.residual.6",
+      "shape": [
+        512,
+        512,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 0,
+      "bias": true,
+      "pure2d": true,
+      "fraction": 0.6
+    },
+    {
+      "path": "upsamples.3.upsamples.0.residual.2",
+      "shape": [
+        256,
+        512,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 0,
+      "bias": true,
+      "pure2d": true,
+      "fraction": 0.6
+    },
+    {
+      "path": "upsamples.3.upsamples.0.residual.6",
+      "shape": [
+        256,
+        256,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 0,
+      "bias": true,
+      "pure2d": true,
+      "fraction": 0.6
+    },
+    {
+      "path": "upsamples.3.upsamples.1.residual.2",
+      "shape": [
+        256,
+        256,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 0,
+      "bias": true,
+      "pure2d": true,
+      "fraction": 0.6
+    },
+    {
+      "path": "upsamples.3.upsamples.1.residual.6",
+      "shape": [
+        256,
+        256,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 0,
+      "bias": true,
+      "pure2d": true,
+      "fraction": 0.6
+    },
+    {
+      "path": "upsamples.3.upsamples.2.residual.2",
+      "shape": [
+        256,
+        256,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 0,
+      "bias": true,
+      "pure2d": true,
+      "fraction": 0.6
+    },
+    {
+      "path": "upsamples.3.upsamples.2.residual.6",
+      "shape": [
+        256,
+        256,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 0,
+      "bias": true,
+      "pure2d": true,
+      "fraction": 0.6
+    },
+    {
+      "path": "head.2",
+      "shape": [
+        12,
+        256,
+        3,
+        3,
+        3
+      ],
+      "padding": 1,
+      "rank": 7,
+      "bias": true,
+      "pure2d": false,
+      "fraction": 0.6
+    }
+  ],
+  "provenance": {
+    "source": "/data/nvme6/gushiqiao/codes/latest/vae/LightX2V/lightx2v_train/experiments/wan22_temporal_search/runs/full_depth_sum60_tail2d_reconstruction5k_gpu0123/recover/checkpoint-005000.pt",
+    "source_step": 5000,
+    "source_sha256": "6244ce47cb45d3e08bf03c71bb758082567d53f17e2de046fd9329aa79ece423",
+    "source_bytes": 4153786363
+  },
+  "step": 300
+}

lightvae-pro-wan22-decoder.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9985d3b90321f48109c3fed25b526f38d137a1a7a8f34e1afbe41ddfc27d2782
+size 1384550888

manifest.json

@@ -0,0 +1,124 @@
+{
+  "format_version": 1,
+  "created_utc": "2026-10-10T06:06:17.467435+00:00",
+  "source_files_unchanged": true,
+  "precision_policy": "Preserve all tensor names, shapes, dtypes and values.",
+  "loading_notes": {
+    "wan": "Use safetensors.torch.load_file for the model state. Original checkpoint fields (family, kept, packed_records, provenance, step, etc.) are preserved in the checkpoint_metadata header JSON and sidecar. The existing torch.load-based .pt loader cannot load these files unchanged.",
+    "h3": "Byte-for-byte copy of the original component checkpoint, including header metadata.",
+    "components": "Decoder files do not include an encoder. The H3 encoder is pruned, not original.",
+    "wan21_lite_source": "Uses wan21_compact_latest.pt as requested, not the later recovery/student.pt."
+  },
+  "weights": [
+    {
+      "file": "lightvae-pro-wan21-decoder.safetensors",
+      "metadata_file": "lightvae-pro-wan21-decoder.metadata.json",
+      "source": "/data/nvme6/gushiqiao/codes/latest/vae_prune/models/wan21_full_depth_latest.pt",
+      "source_sha256": "6411aa9c9c8d4498b962dffafc738def3b168c6a4f6bcd793d3a10d6c3ed01c1",
+      "sha256": "15597278a84b3b416f5f41552ed38a6145a403e5792760979f7bb74f7fb5c0eb",
+      "size_bytes": 179047668,
+      "validation": "all_tensor_keys_shapes_dtypes_and_bytes_equal; metadata_equal",
+      "tensor_count": 130,
+      "tensor_prefix_counts": {
+        "conv2": 2,
+        "decoder": 126,
+        "latents_mean": 1,
+        "latents_std": 1
+      },
+      "tensor_dtype_counts": {
+        "F32": 130
+      }
+    },
+    {
+      "file": "lightvae-lite-wan21-decoder.safetensors",
+      "metadata_file": "lightvae-lite-wan21-decoder.metadata.json",
+      "source": "/data/nvme6/gushiqiao/codes/latest/vae_prune/models/wan21_compact_latest.pt",
+      "source_sha256": "18c07c937083b39c768c033d9da91a39e02713072b900b2f8d7a00bca313b659",
+      "sha256": "05cbfff899d1bd8ce5afa0cc5ed03dcd01dd95876a76b1fb7df5427353f822ce",
+      "size_bytes": 70875748,
+      "validation": "all_tensor_keys_shapes_dtypes_and_bytes_equal; metadata_equal",
+      "tensor_count": 66,
+      "tensor_prefix_counts": {
+        "conv2": 2,
+        "decoder": 62,
+        "latents_mean": 1,
+        "latents_std": 1
+      },
+      "tensor_dtype_counts": {
+        "F32": 66
+      }
+    },
+    {
+      "file": "lightvae-pro-wan22-decoder.safetensors",
+      "metadata_file": "lightvae-pro-wan22-decoder.metadata.json",
+      "source": "/data/nvme6/gushiqiao/codes/latest/vae_prune/models/wan22_full_depth_latest.pt",
+      "source_sha256": "566bac95f1bbaa413c3a328e3781ec4f858a3c52eddb75a6f2040aa5386e150b",
+      "sha256": "9985d3b90321f48109c3fed25b526f38d137a1a7a8f34e1afbe41ddfc27d2782",
+      "size_bytes": 1384550888,
+      "validation": "all_tensor_keys_shapes_dtypes_and_bytes_equal; metadata_equal",
+      "tensor_count": 130,
+      "tensor_prefix_counts": {
+        "conv2": 2,
+        "decoder": 128
+      },
+      "tensor_dtype_counts": {
+        "F32": 130
+      }
+    },
+    {
+      "file": "lightvae-lite-wan22-decoder.safetensors",
+      "metadata_file": "lightvae-lite-wan22-decoder.metadata.json",
+      "source": "/data/nvme6/gushiqiao/codes/latest/vae_prune/models/wan22_compact_latest.pt",
+      "source_sha256": "a381b74119cbf1b90cd879cad7c587c1a0a2822af730cb67fc5e53f0388215ee",
+      "sha256": "455846ab42fe580b82209aec2bee84774860e29a6a66673ec6effe99400b6197",
+      "size_bytes": 609707592,
+      "validation": "all_tensor_keys_shapes_dtypes_and_bytes_equal; metadata_equal",
+      "tensor_count": 66,
+      "tensor_prefix_counts": {
+        "conv2": 2,
+        "decoder": 64
+      },
+      "tensor_dtype_counts": {
+        "F32": 66
+      }
+    },
+    {
+      "file": "lightvae-lite-h3-decoder.safetensors",
+      "metadata_file": "lightvae-lite-h3-decoder.metadata.json",
+      "source": "/data/nvme6/gushiqiao/codes/latest/vae_prune/models/minimax_h3_lightvae_decoder.safetensors",
+      "source_sha256": "d060ad616a33f422fcb8395f525d6d71185b5ddb3e016b31cc210ba2c9c06ee9",
+      "sha256": "d060ad616a33f422fcb8395f525d6d71185b5ddb3e016b31cc210ba2c9c06ee9",
+      "size_bytes": 3248356840,
+      "validation": "whole_file_sha256_equal_including_header_metadata",
+      "tensor_count": 203,
+      "tensor_prefix_counts": {
+        "decoder": 199,
+        "latents_mean": 1,
+        "latents_std": 1,
+        "post_quant_conv": 2
+      },
+      "tensor_dtype_counts": {
+        "F32": 203
+      }
+    },
+    {
+      "file": "lightvae-lite-h3-encoder.safetensors",
+      "metadata_file": "lightvae-lite-h3-encoder.metadata.json",
+      "source": "/data/nvme6/gushiqiao/codes/latest/vae_prune/models/minimax_h3_lightvae_encoder.safetensors",
+      "source_sha256": "4b0e5af102da19324c64df40abbd95008d8acb0b9cf76e4704d74c8ac20cefac",
+      "sha256": "4b0e5af102da19324c64df40abbd95008d8acb0b9cf76e4704d74c8ac20cefac",
+      "size_bytes": 406335344,
+      "validation": "whole_file_sha256_equal_including_header_metadata",
+      "tensor_count": 72,
+      "tensor_prefix_counts": {
+        "encoder": 68,
+        "latents_mean": 1,
+        "latents_std": 1,
+        "quant_conv": 2
+      },
+      "tensor_dtype_counts": {
+        "F32": 72
+      }
+    }
+  ]
+}