Add files using upload-large-folder tool

README.md

@@ -0,0 +1,95 @@
+---
+license: other
+tags: [video-generation, text-to-audio-video, reference-to-video, minimax-h3, fastvideo, fp8]
+---
+
+# FastH3-Omni-FP8 (OmniRef, FP8 W8A8 DiT, for GPUs without FP4)
+
+FastH3 OmniRef (reference-to-video with audio, 8-forward PDD) as an FP8 stack for GPUs that have no FP4 tensor cores:
+RTX 4090 / L40S / RTX 6000 Ada (sm_89), H100 / H200 (sm_90), and any Blackwell part when you prefer FP8.
+It is the original distilled OmniRef student (step 3500) quantized to FP8; nothing is retrained.
+
+## Which weights, and why
+
+Two bf16 sources were quantized the same way (FP8 e4m3, one scale per output channel, activations per token at run time)
+and scored against the bf16 OmniRef teacher with the QAD harness metric: teacher-forced x0 relative L2 per forward on the
+15 held-out OmniRef rows (480p and 768p), averaged over the 8 forwards; `score` is the mean of the video and audio
+columns. Lower is better.
+
+| Row | Video | Audio | Score |
+|---|---|---|---|
+| bf16 OmniRef step 3500 (the teacher, unquantized) | 0.000 | 0.000 | 0.000 |
+| **(a) step 3500, FP8 W8A8 dynamic (this release)** | **0.094** | **0.168** | **0.131** |
+| (b) NVFP4 QAD v2 masters (step 120), FP8 W8A8 dynamic | 0.131 | 0.268 | 0.200 |
+| (b) NVFP4 QAD v2 masters, unquantized bf16 (drift from the teacher) | 0.119 | 0.249 | 0.184 |
+| (a) FP8 weights only, bf16 activations | 0.084 | 0.136 | 0.110 |
+| (c) (a) with static per-tensor activation scales, 1000-row max calibration | 0.098 | 0.171 | 0.134 |
+| (c) (a) with static scales, MSE-clipped on 24 rows | 0.096 | 0.172 | 0.134 |
+| for context: NVFP4 initialization (step 0 of the QAD runs) | 0.225 | 0.499 | 0.362 |
+| for context: NVFP4 QAD v2, step 120 | 0.198 | 0.409 | 0.303 |
+
+Findings:
+
+- **The original student quantizes better to FP8 than the QAD v2 masters.** QAD moved the weights 0.18 away from the
+  teacher in bf16 (it tuned them to absorb 4-bit rounding and FP4 attention); FP8's much finer rounding gets nothing back
+  from that, so (b) stays 0.18 off the teacher before any FP8 rounding. (a) is used.
+- FP8 error is mostly in the weights (0.110 of the 0.131); activation rounding adds the rest.
+- **Static activation scales do not buy accuracy and FastVideo's FP8 loader does not support them.** With e4m3 the
+  rounding error is relative, so the per-tensor scale hardly matters unless values overflow or underflow: MSE clipping chose
+  the full calibrated amax for almost every layer, and static scales score 0.134 against 0.131 for dynamic per-token. They
+  are not exported. Using them would need a small kernel and loader change (a stored per-layer activation scale, and a
+  quantize step that takes it instead of reducing each row), which would also drop the per-token amax pass.
+- The NVFP4 rows use FP4 sparse attention; every FP8 row uses bf16 attention, as the FP8 profile does. Scores are
+  therefore not an isolated linear-layer comparison, but they are what each stack delivers.
+
+## Contents
+
+| Component | Format |
+|---|---|
+| `transformer` / `transformer_ref` | Identical. 300 linears (q, k, v, out, fc_in, fc_out of all 50 blocks) as `float8_e4m3fn` with a float32 `weight_scale` per output channel (the `FastVideo-FastH3-8-Step-V2-FP8` format); everything else bf16: AdaLN/timestep projections, embeddings, norms, the VSA compression gates, the token refiner |
+| `text_encoder` | NVFP4 Qwen3-VL text encoder (unchanged from the NVFP4 releases; on an RTX 4090 it streams per layer and expands to bf16) |
+| `vae` | Dense light VAE (26-layer decoder), from the V2 Consumer release. Used for reference encoding and decoding; the NVFP4 decoder needs FP4 hardware, so there is none here |
+| `audio_vae`, `scheduler`, `audio_scheduler`, `processor`, `tokenizer`, `fastvideo_inference.json` | Unchanged |
+| `adaln_table.pt` | Precomputed AdaLN modulation for the contract's fixed timestep ladder, image references with and without a reference audio. Lets the loader skip the 24 GiB of `adaln_proj` weights |
+
+Size: DiT 47.4 GiB (24.2 GiB of it is the AdaLN projections the table replaces, 18.0 GiB FP8 linears, 5.2 GiB other),
+encoder 16 GiB, VAE 4 GiB, audio VAE 0.6 GiB, table 0.5 GiB: about 68 GiB on disk with the duplicate `transformer`
+stored once.
+
+## Running with FastVideo
+
+Needs the FastVideo branch with the OmniRef consumer profiles. `--base-model-path` points at this folder so the encoder
+and the light VAE come from it rather than from base MiniMax-H3.
+
+```bash
+python examples/inference/basic/basic_fasth3_omniref_pdd.py \
+  --model-path ./FastH3-Omni-FP8 --base-model-path ./FastH3-Omni-FP8 \
+  --composed-dir ./FastH3-Omni-FP8-composed \
+  --adaln-table ./FastH3-Omni-FP8/adaln_table.pt \
+  --image reference.png --prompt "a person on a sofa, lit by a warm lamp, speaking to camera" \
+  --profile rtx-4090
+```
+
+`--profile rtx-4090` selects FP8 linears (`transformer_quant: FP8`, per-channel weights, per-token activations),
+bf16 VSA attention on the contract's 128-token tiles (`FASTVIDEO_H3_VSA_FINE_ATTENTION=emulator-bf16`), the dense VAE
+decoder, and a non-resident encoder. On Hopper or Blackwell data-center GPUs the same command works with the `rtx-4090`
+profile, or without a profile and with `FASTVIDEO_H3_VSA_FP4=0`.
+
+The AdaLN table was checked with one image reference, an image plus a reference audio, and a reference video plus an
+image. A layout whose row timesteps differ raises "No precomputed AdaLN modulation"; drop `--adaln-table` then (the
+projection weights are in the DiT).
+
+## Checks
+
+- Clips rendered on a GB200 through this folder and `--profile rtx-4090` (480p, 5.2 s, 124 frames, two prompts,
+  one with a reference audio) are clean: no ghosting, tile seams, flicker or audio dropouts, with the same composition
+  as the bf16 student at the same seed.
+- The export loads through the existing pre-quantized FP8 loader ("Loaded 300 pre-quantized FP8 linears").
+- The scoring harness rounds weights and activations with the same math as the export and the deployed kernels, in a
+  bf16 GEMM on dequantized operands; the real `_scaled_mm` path was not scored separately.
+- Not tested on a 4090. Peak GPU memory allocated on the GB200 at 480p with the AdaLN table was 37.9 GiB (one image),
+  38.3 GiB (image plus audio), 48.6 GiB (video plus image reference). The DiT weights alone are about 23 GiB without the
+  AdaLN projections, so a 24 GB card will probably need layerwise block streaming (`offload.dit_layerwise: true` in a YAML
+  recipe; the example script does not expose it).
+
+Internal release candidate. Private; not for redistribution.

adaln_table.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:61bd6dabdac6567e4dece2d169e28bc39724da1da307a0cd967c8247980a1755
+size 522858145

audio_scheduler/scheduler_config.json

@@ -0,0 +1,5 @@
+{
+  "_class_name": "MiniMaxH3Scheduler",
+  "_diffusers_version": "0.36.0.dev0",
+  "shift": 3.0
+}

audio_vae/config.json

@@ -0,0 +1,125 @@
+{
+  "_class_name": "AutoencoderKLMiniMaxH3Audio",
+  "_diffusers_version": "0.36.0.dev0",
+  "encoder_dim": 64,
+  "encoder_rates": [
+    2,
+    4,
+    4,
+    5,
+    5
+  ],
+  "latent_dim": 2048,
+  "latent_channels": 32,
+  "decoder_dim": 1024,
+  "decoder_rates": [
+    5,
+    5,
+    2,
+    2,
+    2,
+    2,
+    2
+  ],
+  "decoder_kernel_sizes": [
+    9,
+    9,
+    4,
+    4,
+    4,
+    4,
+    4
+  ],
+  "num_attention_heads": 8,
+  "resblock_kernel_sizes": [
+    3,
+    7,
+    11
+  ],
+  "resblock_dilation_sizes": [
+    [
+      1,
+      3,
+      5
+    ],
+    [
+      1,
+      3,
+      5
+    ],
+    [
+      1,
+      3,
+      5
+    ]
+  ],
+  "sampling_rate": 32000,
+  "latents_mean": [
+    -0.020211687488382354,
+    0.3876466479950502,
+    -0.04398279799186767,
+    -0.28591514936373,
+    0.08179686214561671,
+    -0.35782641352446604,
+    0.040623809960919084,
+    -0.01552534501956604,
+    -0.223362481667332,
+    0.1821006842509091,
+    0.2941778783780663,
+    -0.07901167601970885,
+    -0.056815072777201,
+    -0.3699028221860095,
+    -0.31616315591624855,
+    0.5905951377425391,
+    -0.052139568068853864,
+    0.013673160263486295,
+    -0.03691647864630577,
+    0.09732660653298163,
+    -0.3394662328788498,
+    -0.30685677538541667,
+    -0.24504598907458763,
+    -0.034698524462007344,
+    0.02868032184767538,
+    -0.21217779266454084,
+    -0.1678263169941987,
+    0.3221287889040614,
+    -0.1223055851554907,
+    0.4356604928128464,
+    -0.0502599202236253,
+    0.3979258376211797
+  ],
+  "latents_std": [
+    1.6895524230479284,
+    2.76263727217653,
+    1.7945344281264435,
+    1.6801681847309828,
+    1.6390226546605453,
+    2.7788298348882177,
+    1.7659090095747236,
+    1.6199757612137327,
+    2.6336525640336896,
+    1.8539356672817833,
+    2.5056497896915633,
+    1.811019237886178,
+    1.9579657790720237,
+    1.6685498243529284,
+    1.4922469314453364,
+    3.298670198067373,
+    1.9491804496832168,
+    1.8720003270431442,
+    1.8334080103291832,
+    1.6488070416529093,
+    1.6176957696319716,
+    1.9131449234774398,
+    1.5695245398428617,
+    1.6943659940415912,
+    1.8318420762504692,
+    1.5540637421583379,
+    1.9344930328968526,
+    1.599198216109855,
+    1.718045989838149,
+    1.6307219190837705,
+    1.8661226051202384,
+    1.5613768203168363
+  ]
+}

audio_vae/diffusion_pytorch_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:52c59e67ba8de5477c81bfbced0327aabf500f1bfdeefd5ee754529241cb26cb
+size 605429340

fastvideo_inference.json

@@ -0,0 +1,31 @@
+{
+  "attention_backend": "VIDEO_SPARSE_ATTN_H3",
+  "audio_scheduler_shift": 3.0,
+  "base_model_revision": "hf://MiniMaxAI/MiniMax-H3@9bfb6693f2cf6de171db46d1aa586f67d773a1da",
+  "conditioning": "fixed_ordered_references_target_only_flow",
+  "grid_max_t": 0.999,
+  "guidance_scale": 1.0,
+  "model_type": "ref2va",
+  "num_inference_steps": 8,
+  "pdd_step_indices": [
+    0,
+    4,
+    8,
+    12,
+    16,
+    20,
+    24,
+    28,
+    32
+  ],
+  "pdd_steps": 32,
+  "schema": "fasth3-inference-contract-v1",
+  "schema_version": "fasth3-inference-contract-v1",
+  "transformer_component": "transformer_ref",
+  "transformer_forwards": 8,
+  "video_scheduler_shift": 12.0,
+  "vsa_ref_keep_rate": 0.1,
+  "vsa_ref_policy": "p2_multi_region",
+  "vsa_sparsity": 0.9,
+  "vsa_tile_size": 128
+}

modular_model_index.json

@@ -0,0 +1,131 @@
+{
+  "_class_name": "MiniMaxH3ModularPipeline",
+  "_diffusers_version": "0.36.0.dev0",
+  "_blocks_class_name": "MiniMaxH3Blocks",
+  "text_encoder": [
+    "transformers",
+    "Qwen3VLForConditionalGeneration",
+    {
+      "type_hint": [
+        "transformers",
+        "Qwen3VLForConditionalGeneration"
+      ],
+      "pretrained_model_name_or_path": "MiniMaxAI/MiniMax-H3",
+      "subfolder": "text_encoder",
+      "variant": null,
+      "revision": null
+    }
+  ],
+  "tokenizer": [
+    "transformers",
+    "Qwen2TokenizerFast",
+    {
+      "type_hint": [
+        "transformers",
+        "Qwen2TokenizerFast"
+      ],
+      "pretrained_model_name_or_path": "MiniMaxAI/MiniMax-H3",
+      "subfolder": "tokenizer",
+      "variant": null,
+      "revision": null
+    }
+  ],
+  "processor": [
+    "transformers",
+    "Qwen3VLProcessor",
+    {
+      "type_hint": [
+        "transformers",
+        "Qwen3VLProcessor"
+      ],
+      "pretrained_model_name_or_path": "MiniMaxAI/MiniMax-H3",
+      "subfolder": "processor",
+      "variant": null,
+      "revision": null
+    }
+  ],
+  "vae": [
+    "diffusers",
+    "AutoencoderKLMiniMaxH3",
+    {
+      "type_hint": [
+        "diffusers",
+        "AutoencoderKLMiniMaxH3"
+      ],
+      "pretrained_model_name_or_path": "MiniMaxAI/MiniMax-H3",
+      "subfolder": "vae",
+      "variant": null,
+      "revision": null
+    }
+  ],
+  "audio_vae": [
+    "diffusers",
+    "AutoencoderKLMiniMaxH3Audio",
+    {
+      "type_hint": [
+        "diffusers",
+        "AutoencoderKLMiniMaxH3Audio"
+      ],
+      "pretrained_model_name_or_path": "MiniMaxAI/MiniMax-H3",
+      "subfolder": "audio_vae",
+      "variant": null,
+      "revision": null
+    }
+  ],
+  "transformer": [
+    "diffusers",
+    "MiniMaxH3Transformer3DModel",
+    {
+      "type_hint": [
+        "diffusers",
+        "MiniMaxH3Transformer3DModel"
+      ],
+      "pretrained_model_name_or_path": "MiniMaxAI/MiniMax-H3",
+      "subfolder": "transformer",
+      "variant": null,
+      "revision": null
+    }
+  ],
+  "transformer_ref": [
+    "diffusers",
+    "MiniMaxH3Transformer3DModel",
+    {
+      "type_hint": [
+        "diffusers",
+        "MiniMaxH3Transformer3DModel"
+      ],
+      "pretrained_model_name_or_path": "MiniMaxAI/MiniMax-H3",
+      "subfolder": "transformer_ref",
+      "variant": null,
+      "revision": null
+    }
+  ],
+  "scheduler": [
+    "diffusers",
+    "MiniMaxH3Scheduler",
+    {
+      "type_hint": [
+        "diffusers",
+        "MiniMaxH3Scheduler"
+      ],
+      "pretrained_model_name_or_path": "MiniMaxAI/MiniMax-H3",
+      "subfolder": "scheduler",
+      "variant": null,
+      "revision": null
+    }
+  ],
+  "audio_scheduler": [
+    "diffusers",
+    "MiniMaxH3Scheduler",
+    {
+      "type_hint": [
+        "diffusers",
+        "MiniMaxH3Scheduler"
+      ],
+      "pretrained_model_name_or_path": "MiniMaxAI/MiniMax-H3",
+      "subfolder": "audio_scheduler",
+      "variant": null,
+      "revision": null
+    }
+  ]
+}

processor/chat_template.json

@@ -0,0 +1,3 @@
+{
+    "chat_template": "{%- if tools %}\n    {{- '<|im_start|>system\\n' }}\n    {%- if messages[0].role == 'system' %}\n        {%- if messages[0].content is string %}\n            {{- messages[0].content }}\n        {%- else %}\n            {%- for content in messages[0].content %}\n                {%- if 'text' in content %}\n                    {{- content.text }}\n                {%- endif %}\n            {%- endfor %}\n        {%- endif %}\n        {{- '\\n\\n' }}\n    {%- endif %}\n    {{- \"# Tools\\n\\nYou may call one or more functions to assist with the user query.\\n\\nYou are provided with function signatures within <tools></tools> XML tags:\\n<tools>\" }}\n    {%- for tool in tools %}\n        {{- \"\\n\" }}\n        {{- tool | tojson }}\n    {%- endfor %}\n    {{- \"\\n</tools>\\n\\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\\n<tool_call>\\n{\\\"name\\\": <function-name>, \\\"arguments\\\": <args-json-object>}\\n</tool_call><|im_end|>\\n\" }}\n{%- else %}\n    {%- if messages[0].role == 'system' %}\n        {{- '<|im_start|>system\\n' }}\n        {%- if messages[0].content is string %}\n            {{- messages[0].content }}\n        {%- else %}\n            {%- for content in messages[0].content %}\n                {%- if 'text' in content %}\n                    {{- content.text }}\n                {%- endif %}\n            {%- endfor %}\n        {%- endif %}\n        {{- '<|im_end|>\\n' }}\n    {%- endif %}\n{%- endif %}\n{%- set image_count = namespace(value=0) %}\n{%- set video_count = namespace(value=0) %}\n{%- for message in messages %}\n    {%- if message.role == \"user\" %}\n        {{- '<|im_start|>' + message.role + '\\n' }}\n        {%- if message.content is string %}\n            {{- message.content }}\n        {%- else %}\n            {%- for content in message.content %}\n                {%- if content.type == 'image' or 'image' in content or 'image_url' in content %}\n                    {%- set image_count.value = image_count.value + 1 %}\n                    {%- if add_vision_id %}Picture {{ image_count.value }}: {% endif -%}\n                    <|vision_start|><|image_pad|><|vision_end|>\n                {%- elif content.type == 'video' or 'video' in content %}\n                    {%- set video_count.value = video_count.value + 1 %}\n                    {%- if add_vision_id %}Video {{ video_count.value }}: {% endif -%}\n                    <|vision_start|><|video_pad|><|vision_end|>\n                {%- elif 'text' in content %}\n                    {{- content.text }}\n                {%- endif %}\n            {%- endfor %}\n        {%- endif %}\n        {{- '<|im_end|>\\n' }}\n    {%- elif message.role == \"assistant\" %}\n        {{- '<|im_start|>' + message.role + '\\n' }}\n        {%- if message.content is string %}\n            {{- message.content }}\n        {%- else %}\n            {%- for content_item in message.content %}\n                {%- if 'text' in content_item %}\n                    {{- content_item.text }}\n                {%- endif %}\n            {%- endfor %}\n        {%- endif %}\n        {%- if message.tool_calls %}\n            {%- for tool_call in message.tool_calls %}\n                {%- if (loop.first and message.content) or (not loop.first) %}\n                    {{- '\\n' }}\n                {%- endif %}\n                {%- if tool_call.function %}\n                    {%- set tool_call = tool_call.function %}\n                {%- endif %}\n                {{- '<tool_call>\\n{\"name\": \"' }}\n                {{- tool_call.name }}\n                {{- '\", \"arguments\": ' }}\n                {%- if tool_call.arguments is string %}\n                    {{- tool_call.arguments }}\n                {%- else %}\n                    {{- tool_call.arguments | tojson }}\n                {%- endif %}\n                {{- '}\\n</tool_call>' }}\n            {%- endfor %}\n        {%- endif %}\n        {{- '<|im_end|>\\n' }}\n    {%- elif message.role == \"tool\" %}\n        {%- if loop.first or (messages[loop.index0 - 1].role != \"tool\") %}\n            {{- '<|im_start|>user' }}\n        {%- endif %}\n        {{- '\\n<tool_response>\\n' }}\n        {%- if message.content is string %}\n            {{- message.content }}\n        {%- else %}\n            {%- for content in message.content %}\n                {%- if content.type == 'image' or 'image' in content or 'image_url' in content %}\n                    {%- set image_count.value = image_count.value + 1 %}\n                    {%- if add_vision_id %}Picture {{ image_count.value }}: {% endif -%}\n                    <|vision_start|><|image_pad|><|vision_end|>\n                {%- elif content.type == 'video' or 'video' in content %}\n                    {%- set video_count.value = video_count.value + 1 %}\n                    {%- if add_vision_id %}Video {{ video_count.value }}: {% endif -%}\n                    <|vision_start|><|video_pad|><|vision_end|>\n                {%- elif 'text' in content %}\n                    {{- content.text }}\n                {%- endif %}\n            {%- endfor %}\n        {%- endif %}\n        {{- '\\n</tool_response>' }}\n        {%- if loop.last or (messages[loop.index0 + 1].role != \"tool\") %}\n            {{- '<|im_end|>\\n' }}\n        {%- endif %}\n    {%- endif %}\n{%- endfor %}\n{%- if add_generation_prompt %}\n    {{- '<|im_start|>assistant\\n' }}\n{%- endif %}\n"
+}

processor/preprocessor_config.json

@@ -0,0 +1,21 @@
+{
+    "size": {
+        "longest_edge": 16777216,
+        "shortest_edge": 65536
+    },
+    "patch_size": 16,
+    "temporal_patch_size": 2,
+    "merge_size": 2,
+    "image_mean": [
+        0.5,
+        0.5,
+        0.5
+    ],
+    "image_std": [
+        0.5,
+        0.5,
+        0.5
+    ],
+    "processor_class": "Qwen3VLProcessor",
+    "image_processor_type": "Qwen2VLImageProcessorFast"
+}

processor/tokenizer_config.json

@@ -0,0 +1,246 @@
+{
+  "add_bos_token": false,
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151646": {
+      "content": "<|object_ref_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151647": {
+      "content": "<|object_ref_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151648": {
+      "content": "<|box_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151649": {
+      "content": "<|box_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151650": {
+      "content": "<|quad_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151651": {
+      "content": "<|quad_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151652": {
+      "content": "<|vision_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151653": {
+      "content": "<|vision_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151654": {
+      "content": "<|vision_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151655": {
+      "content": "<|image_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151656": {
+      "content": "<|video_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151657": {
+      "content": "<tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151658": {
+      "content": "</tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151659": {
+      "content": "<|fim_prefix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151660": {
+      "content": "<|fim_middle|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151661": {
+      "content": "<|fim_suffix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151662": {
+      "content": "<|fim_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151663": {
+      "content": "<|repo_name|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151664": {
+      "content": "<|file_sep|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151665": {
+      "content": "<tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151666": {
+      "content": "</tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151667": {
+      "content": "<think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151668": {
+      "content": "</think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>",
+    "<d>",
+    "</d>",
+    "<|cutoff|>",
+    "<|lyrics_start|>",
+    "<|lyrics_end|>",
+    "<|caption_start|>",
+    "<|caption_end|>"
+  ],
+  "bos_token": null,
+  "chat_template": "{%- if tools %}\n    {{- '<|im_start|>system\\n' }}\n    {%- if messages[0].role == 'system' %}\n        {%- if messages[0].content is string %}\n            {{- messages[0].content }}\n        {%- else %}\n            {%- for content in messages[0].content %}\n                {%- if 'text' in content %}\n                    {{- content.text }}\n                {%- endif %}\n            {%- endfor %}\n        {%- endif %}\n        {{- '\\n\\n' }}\n    {%- endif %}\n    {{- \"# Tools\\n\\nYou may call one or more functions to assist with the user query.\\n\\nYou are provided with function signatures within <tools></tools> XML tags:\\n<tools>\" }}\n    {%- for tool in tools %}\n        {{- \"\\n\" }}\n        {{- tool | tojson }}\n    {%- endfor %}\n    {{- \"\\n</tools>\\n\\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\\n<tool_call>\\n{\\\"name\\\": <function-name>, \\\"arguments\\\": <args-json-object>}\\n</tool_call><|im_end|>\\n\" }}\n{%- else %}\n    {%- if messages[0].role == 'system' %}\n        {{- '<|im_start|>system\\n' }}\n        {%- if messages[0].content is string %}\n            {{- messages[0].content }}\n        {%- else %}\n            {%- for content in messages[0].content %}\n                {%- if 'text' in content %}\n                    {{- content.text }}\n                {%- endif %}\n            {%- endfor %}\n        {%- endif %}\n        {{- '<|im_end|>\\n' }}\n    {%- endif %}\n{%- endif %}\n{%- set image_count = namespace(value=0) %}\n{%- set video_count = namespace(value=0) %}\n{%- for message in messages %}\n    {%- if message.role == \"user\" %}\n        {{- '<|im_start|>' + message.role + '\\n' }}\n        {%- if message.content is string %}\n            {{- message.content }}\n        {%- else %}\n            {%- for content in message.content %}\n                {%- if content.type == 'image' or 'image' in content or 'image_url' in content %}\n                    {%- set image_count.value = image_count.value + 1 %}\n                    {%- if add_vision_id %}Picture {{ image_count.value }}: {% endif -%}\n                    <|vision_start|><|image_pad|><|vision_end|>\n                {%- elif content.type == 'video' or 'video' in content %}\n                    {%- set video_count.value = video_count.value + 1 %}\n                    {%- if add_vision_id %}Video {{ video_count.value }}: {% endif -%}\n                    <|vision_start|><|video_pad|><|vision_end|>\n                {%- elif 'text' in content %}\n                    {{- content.text }}\n                {%- endif %}\n            {%- endfor %}\n        {%- endif %}\n        {{- '<|im_end|>\\n' }}\n    {%- elif message.role == \"assistant\" %}\n        {{- '<|im_start|>' + message.role + '\\n' }}\n        {%- if message.content is string %}\n            {{- message.content }}\n        {%- else %}\n            {%- for content_item in message.content %}\n                {%- if 'text' in content_item %}\n                    {{- content_item.text }}\n                {%- endif %}\n            {%- endfor %}\n        {%- endif %}\n        {%- if message.tool_calls %}\n            {%- for tool_call in message.tool_calls %}\n                {%- if (loop.first and message.content) or (not loop.first) %}\n                    {{- '\\n' }}\n                {%- endif %}\n                {%- if tool_call.function %}\n                    {%- set tool_call = tool_call.function %}\n                {%- endif %}\n                {{- '<tool_call>\\n{\"name\": \"' }}\n                {{- tool_call.name }}\n                {{- '\", \"arguments\": ' }}\n                {%- if tool_call.arguments is string %}\n                    {{- tool_call.arguments }}\n                {%- else %}\n                    {{- tool_call.arguments | tojson }}\n                {%- endif %}\n                {{- '}\\n</tool_call>' }}\n            {%- endfor %}\n        {%- endif %}\n        {{- '<|im_end|>\\n' }}\n    {%- elif message.role == \"tool\" %}\n        {%- if loop.first or (messages[loop.index0 - 1].role != \"tool\") %}\n            {{- '<|im_start|>user' }}\n        {%- endif %}\n        {{- '\\n<tool_response>\\n' }}\n        {%- if message.content is string %}\n            {{- message.content }}\n        {%- else %}\n            {%- for content in message.content %}\n                {%- if content.type == 'image' or 'image' in content or 'image_url' in content %}\n                    {%- set image_count.value = image_count.value + 1 %}\n                    {%- if add_vision_id %}Picture {{ image_count.value }}: {% endif -%}\n                    <|vision_start|><|image_pad|><|vision_end|>\n                {%- elif content.type == 'video' or 'video' in content %}\n                    {%- set video_count.value = video_count.value + 1 %}\n                    {%- if add_vision_id %}Video {{ video_count.value }}: {% endif -%}\n                    <|vision_start|><|video_pad|><|vision_end|>\n                {%- elif 'text' in content %}\n                    {{- content.text }}\n                {%- endif %}\n            {%- endfor %}\n        {%- endif %}\n        {{- '\\n</tool_response>' }}\n        {%- if loop.last or (messages[loop.index0 + 1].role != \"tool\") %}\n            {{- '<|im_end|>\\n' }}\n        {%- endif %}\n    {%- endif %}\n{%- endfor %}\n{%- if add_generation_prompt %}\n    {{- '<|im_start|>assistant\\n' }}\n{%- endif %}\n",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "model_max_length": 262144,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

processor/video_preprocessor_config.json

@@ -0,0 +1,21 @@
+{
+    "size": {
+        "longest_edge": 25165824,
+        "shortest_edge": 4096
+    },
+    "patch_size": 16,
+    "temporal_patch_size": 2,
+    "merge_size": 2,
+    "image_mean": [
+        0.5,
+        0.5,
+        0.5
+    ],
+    "image_std": [
+        0.5,
+        0.5,
+        0.5
+    ],
+    "processor_class": "Qwen3VLProcessor",
+    "video_processor_type": "Qwen3VLVideoProcessor"
+}

scheduler/scheduler_config.json

@@ -0,0 +1,5 @@
+{
+  "_class_name": "MiniMaxH3Scheduler",
+  "_diffusers_version": "0.36.0.dev0",
+  "shift": 12.0
+}

text_encoder/chat_template.json

@@ -0,0 +1,3 @@
+{
+    "chat_template": "{%- if tools %}\n    {{- '<|im_start|>system\\n' }}\n    {%- if messages[0].role == 'system' %}\n        {%- if messages[0].content is string %}\n            {{- messages[0].content }}\n        {%- else %}\n            {%- for content in messages[0].content %}\n                {%- if 'text' in content %}\n                    {{- content.text }}\n                {%- endif %}\n            {%- endfor %}\n        {%- endif %}\n        {{- '\\n\\n' }}\n    {%- endif %}\n    {{- \"# Tools\\n\\nYou may call one or more functions to assist with the user query.\\n\\nYou are provided with function signatures within <tools></tools> XML tags:\\n<tools>\" }}\n    {%- for tool in tools %}\n        {{- \"\\n\" }}\n        {{- tool | tojson }}\n    {%- endfor %}\n    {{- \"\\n</tools>\\n\\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\\n<tool_call>\\n{\\\"name\\\": <function-name>, \\\"arguments\\\": <args-json-object>}\\n</tool_call><|im_end|>\\n\" }}\n{%- else %}\n    {%- if messages[0].role == 'system' %}\n        {{- '<|im_start|>system\\n' }}\n        {%- if messages[0].content is string %}\n            {{- messages[0].content }}\n        {%- else %}\n            {%- for content in messages[0].content %}\n                {%- if 'text' in content %}\n                    {{- content.text }}\n                {%- endif %}\n            {%- endfor %}\n        {%- endif %}\n        {{- '<|im_end|>\\n' }}\n    {%- endif %}\n{%- endif %}\n{%- set image_count = namespace(value=0) %}\n{%- set video_count = namespace(value=0) %}\n{%- for message in messages %}\n    {%- if message.role == \"user\" %}\n        {{- '<|im_start|>' + message.role + '\\n' }}\n        {%- if message.content is string %}\n            {{- message.content }}\n        {%- else %}\n            {%- for content in message.content %}\n                {%- if content.type == 'image' or 'image' in content or 'image_url' in content %}\n                    {%- set image_count.value = image_count.value + 1 %}\n                    {%- if add_vision_id %}Picture {{ image_count.value }}: {% endif -%}\n                    <|vision_start|><|image_pad|><|vision_end|>\n                {%- elif content.type == 'video' or 'video' in content %}\n                    {%- set video_count.value = video_count.value + 1 %}\n                    {%- if add_vision_id %}Video {{ video_count.value }}: {% endif -%}\n                    <|vision_start|><|video_pad|><|vision_end|>\n                {%- elif 'text' in content %}\n                    {{- content.text }}\n                {%- endif %}\n            {%- endfor %}\n        {%- endif %}\n        {{- '<|im_end|>\\n' }}\n    {%- elif message.role == \"assistant\" %}\n        {{- '<|im_start|>' + message.role + '\\n' }}\n        {%- if message.content is string %}\n            {{- message.content }}\n        {%- else %}\n            {%- for content_item in message.content %}\n                {%- if 'text' in content_item %}\n                    {{- content_item.text }}\n                {%- endif %}\n            {%- endfor %}\n        {%- endif %}\n        {%- if message.tool_calls %}\n            {%- for tool_call in message.tool_calls %}\n                {%- if (loop.first and message.content) or (not loop.first) %}\n                    {{- '\\n' }}\n                {%- endif %}\n                {%- if tool_call.function %}\n                    {%- set tool_call = tool_call.function %}\n                {%- endif %}\n                {{- '<tool_call>\\n{\"name\": \"' }}\n                {{- tool_call.name }}\n                {{- '\", \"arguments\": ' }}\n                {%- if tool_call.arguments is string %}\n                    {{- tool_call.arguments }}\n                {%- else %}\n                    {{- tool_call.arguments | tojson }}\n                {%- endif %}\n                {{- '}\\n</tool_call>' }}\n            {%- endfor %}\n        {%- endif %}\n        {{- '<|im_end|>\\n' }}\n    {%- elif message.role == \"tool\" %}\n        {%- if loop.first or (messages[loop.index0 - 1].role != \"tool\") %}\n            {{- '<|im_start|>user' }}\n        {%- endif %}\n        {{- '\\n<tool_response>\\n' }}\n        {%- if message.content is string %}\n            {{- message.content }}\n        {%- else %}\n            {%- for content in message.content %}\n                {%- if content.type == 'image' or 'image' in content or 'image_url' in content %}\n                    {%- set image_count.value = image_count.value + 1 %}\n                    {%- if add_vision_id %}Picture {{ image_count.value }}: {% endif -%}\n                    <|vision_start|><|image_pad|><|vision_end|>\n                {%- elif content.type == 'video' or 'video' in content %}\n                    {%- set video_count.value = video_count.value + 1 %}\n                    {%- if add_vision_id %}Video {{ video_count.value }}: {% endif -%}\n                    <|vision_start|><|video_pad|><|vision_end|>\n                {%- elif 'text' in content %}\n                    {{- content.text }}\n                {%- endif %}\n            {%- endfor %}\n        {%- endif %}\n        {{- '\\n</tool_response>' }}\n        {%- if loop.last or (messages[loop.index0 + 1].role != \"tool\") %}\n            {{- '<|im_end|>\\n' }}\n        {%- endif %}\n    {%- endif %}\n{%- endfor %}\n{%- if add_generation_prompt %}\n    {{- '<|im_start|>assistant\\n' }}\n{%- endif %}\n"
+}

text_encoder/config.json

@@ -0,0 +1,80 @@
+{
+  "architectures": [
+    "Qwen3VLForConditionalGeneration"
+  ],
+  "image_token_id": 151655,
+  "model_type": "qwen3_vl",
+  "text_config": {
+    "attention_bias": false,
+    "attention_dropout": 0.0,
+    "bos_token_id": 151643,
+    "dtype": "bfloat16",
+    "eos_token_id": 151645,
+    "head_dim": 128,
+    "hidden_act": "silu",
+    "hidden_size": 5120,
+    "initializer_range": 0.02,
+    "intermediate_size": 25600,
+    "max_position_embeddings": 262144,
+    "model_type": "qwen3_vl_text",
+    "num_attention_heads": 64,
+    "num_hidden_layers": 64,
+    "num_key_value_heads": 8,
+    "rms_norm_eps": 1e-06,
+    "rope_scaling": {
+      "mrope_interleaved": true,
+      "mrope_section": [
+        24,
+        20,
+        20
+      ],
+      "rope_type": "default"
+    },
+    "rope_theta": 5000000,
+    "use_cache": true,
+    "vocab_size": 151936
+  },
+  "tie_word_embeddings": false,
+  "transformers_version": "4.57.0.dev0",
+  "video_token_id": 151656,
+  "vision_config": {
+    "deepstack_visual_indexes": [
+      8,
+      16,
+      24
+    ],
+    "depth": 27,
+    "hidden_act": "gelu_pytorch_tanh",
+    "hidden_size": 1152,
+    "in_channels": 3,
+    "initializer_range": 0.02,
+    "intermediate_size": 4304,
+    "model_type": "qwen3_vl",
+    "num_heads": 16,
+    "num_position_embeddings": 2304,
+    "out_hidden_size": 5120,
+    "patch_size": 16,
+    "spatial_merge_size": 2,
+    "temporal_patch_size": 2
+  },
+  "vision_end_token_id": 151653,
+  "vision_start_token_id": 151652,
+  "num_hidden_layers_override": 50,
+  "quantization_config": {
+    "quant_method": "nvfp4",
+    "activation_scheme": "dynamic",
+    "fmt": "e2m1",
+    "group_size": 16,
+    "scale_fmt": "e4m3",
+    "scale_layout": "128x4",
+    "modules_to_not_convert": [
+      "model.visual",
+      "lm_head"
+    ],
+    "producer": {
+      "converter": "convert_minimax_h3_text_encoder_nvfp4.py",
+      "flashinfer": "0.7.1rc2",
+      "kept_language_layers": 50
+    }
+  }
+}

text_encoder/model-00001-of-00004.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:969e7d251a873d16d3f487d9a5e82f8901132cfac9344530aa4e25ba9338e9e7
+size 4290567132

text_encoder/model-00002-of-00004.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ff092cd64c489ce4609a08ba5d9f8548591bf835cd98fa4154a7eccd0ea16dfd
+size 4270035088

text_encoder/model-00003-of-00004.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3dab4af0342cfc9b31839458d059327c7cd1e5be1fe779ded253939cabb36a51
+size 4241221712

text_encoder/model-00004-of-00004.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:628852eb94bd2e3c297d120683f900b49051f236e2cd9d1f26476c30cedc2add
+size 3659204564

text_encoder/preprocessor_config.json

@@ -0,0 +1,21 @@
+{
+    "size": {
+        "longest_edge": 16777216,
+        "shortest_edge": 65536
+    },
+    "patch_size": 16,
+    "temporal_patch_size": 2,
+    "merge_size": 2,
+    "image_mean": [
+        0.5,
+        0.5,
+        0.5
+    ],
+    "image_std": [
+        0.5,
+        0.5,
+        0.5
+    ],
+    "processor_class": "Qwen3VLProcessor",
+    "image_processor_type": "Qwen2VLImageProcessorFast"
+}

text_encoder/tokenizer_config.json

@@ -0,0 +1,246 @@
+{
+  "add_bos_token": false,
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151646": {
+      "content": "<|object_ref_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151647": {
+      "content": "<|object_ref_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151648": {
+      "content": "<|box_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151649": {
+      "content": "<|box_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151650": {
+      "content": "<|quad_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151651": {
+      "content": "<|quad_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151652": {
+      "content": "<|vision_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151653": {
+      "content": "<|vision_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151654": {
+      "content": "<|vision_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151655": {
+      "content": "<|image_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151656": {
+      "content": "<|video_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151657": {
+      "content": "<tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151658": {
+      "content": "</tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151659": {
+      "content": "<|fim_prefix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151660": {
+      "content": "<|fim_middle|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151661": {
+      "content": "<|fim_suffix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151662": {
+      "content": "<|fim_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151663": {
+      "content": "<|repo_name|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151664": {
+      "content": "<|file_sep|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151665": {
+      "content": "<tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151666": {
+      "content": "</tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151667": {
+      "content": "<think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151668": {
+      "content": "</think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>",
+    "<d>",
+    "</d>",
+    "<|cutoff|>",
+    "<|lyrics_start|>",
+    "<|lyrics_end|>",
+    "<|caption_start|>",
+    "<|caption_end|>"
+  ],
+  "bos_token": null,
+  "chat_template": "{%- if tools %}\n    {{- '<|im_start|>system\\n' }}\n    {%- if messages[0].role == 'system' %}\n        {%- if messages[0].content is string %}\n            {{- messages[0].content }}\n        {%- else %}\n            {%- for content in messages[0].content %}\n                {%- if 'text' in content %}\n                    {{- content.text }}\n                {%- endif %}\n            {%- endfor %}\n        {%- endif %}\n        {{- '\\n\\n' }}\n    {%- endif %}\n    {{- \"# Tools\\n\\nYou may call one or more functions to assist with the user query.\\n\\nYou are provided with function signatures within <tools></tools> XML tags:\\n<tools>\" }}\n    {%- for tool in tools %}\n        {{- \"\\n\" }}\n        {{- tool | tojson }}\n    {%- endfor %}\n    {{- \"\\n</tools>\\n\\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\\n<tool_call>\\n{\\\"name\\\": <function-name>, \\\"arguments\\\": <args-json-object>}\\n</tool_call><|im_end|>\\n\" }}\n{%- else %}\n    {%- if messages[0].role == 'system' %}\n        {{- '<|im_start|>system\\n' }}\n        {%- if messages[0].content is string %}\n            {{- messages[0].content }}\n        {%- else %}\n            {%- for content in messages[0].content %}\n                {%- if 'text' in content %}\n                    {{- content.text }}\n                {%- endif %}\n            {%- endfor %}\n        {%- endif %}\n        {{- '<|im_end|>\\n' }}\n    {%- endif %}\n{%- endif %}\n{%- set image_count = namespace(value=0) %}\n{%- set video_count = namespace(value=0) %}\n{%- for message in messages %}\n    {%- if message.role == \"user\" %}\n        {{- '<|im_start|>' + message.role + '\\n' }}\n        {%- if message.content is string %}\n            {{- message.content }}\n        {%- else %}\n            {%- for content in message.content %}\n                {%- if content.type == 'image' or 'image' in content or 'image_url' in content %}\n                    {%- set image_count.value = image_count.value + 1 %}\n                    {%- if add_vision_id %}Picture {{ image_count.value }}: {% endif -%}\n                    <|vision_start|><|image_pad|><|vision_end|>\n                {%- elif content.type == 'video' or 'video' in content %}\n                    {%- set video_count.value = video_count.value + 1 %}\n                    {%- if add_vision_id %}Video {{ video_count.value }}: {% endif -%}\n                    <|vision_start|><|video_pad|><|vision_end|>\n                {%- elif 'text' in content %}\n                    {{- content.text }}\n                {%- endif %}\n            {%- endfor %}\n        {%- endif %}\n        {{- '<|im_end|>\\n' }}\n    {%- elif message.role == \"assistant\" %}\n        {{- '<|im_start|>' + message.role + '\\n' }}\n        {%- if message.content is string %}\n            {{- message.content }}\n        {%- else %}\n            {%- for content_item in message.content %}\n                {%- if 'text' in content_item %}\n                    {{- content_item.text }}\n                {%- endif %}\n            {%- endfor %}\n        {%- endif %}\n        {%- if message.tool_calls %}\n            {%- for tool_call in message.tool_calls %}\n                {%- if (loop.first and message.content) or (not loop.first) %}\n                    {{- '\\n' }}\n                {%- endif %}\n                {%- if tool_call.function %}\n                    {%- set tool_call = tool_call.function %}\n                {%- endif %}\n                {{- '<tool_call>\\n{\"name\": \"' }}\n                {{- tool_call.name }}\n                {{- '\", \"arguments\": ' }}\n                {%- if tool_call.arguments is string %}\n                    {{- tool_call.arguments }}\n                {%- else %}\n                    {{- tool_call.arguments | tojson }}\n                {%- endif %}\n                {{- '}\\n</tool_call>' }}\n            {%- endfor %}\n        {%- endif %}\n        {{- '<|im_end|>\\n' }}\n    {%- elif message.role == \"tool\" %}\n        {%- if loop.first or (messages[loop.index0 - 1].role != \"tool\") %}\n            {{- '<|im_start|>user' }}\n        {%- endif %}\n        {{- '\\n<tool_response>\\n' }}\n        {%- if message.content is string %}\n            {{- message.content }}\n        {%- else %}\n            {%- for content in message.content %}\n                {%- if content.type == 'image' or 'image' in content or 'image_url' in content %}\n                    {%- set image_count.value = image_count.value + 1 %}\n                    {%- if add_vision_id %}Picture {{ image_count.value }}: {% endif -%}\n                    <|vision_start|><|image_pad|><|vision_end|>\n                {%- elif content.type == 'video' or 'video' in content %}\n                    {%- set video_count.value = video_count.value + 1 %}\n                    {%- if add_vision_id %}Video {{ video_count.value }}: {% endif -%}\n                    <|vision_start|><|video_pad|><|vision_end|>\n                {%- elif 'text' in content %}\n                    {{- content.text }}\n                {%- endif %}\n            {%- endfor %}\n        {%- endif %}\n        {{- '\\n</tool_response>' }}\n        {%- if loop.last or (messages[loop.index0 + 1].role != \"tool\") %}\n            {{- '<|im_end|>\\n' }}\n        {%- endif %}\n    {%- endif %}\n{%- endfor %}\n{%- if add_generation_prompt %}\n    {{- '<|im_start|>assistant\\n' }}\n{%- endif %}\n",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "model_max_length": 262144,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

text_encoder/video_preprocessor_config.json

@@ -0,0 +1,21 @@
+{
+    "size": {
+        "longest_edge": 25165824,
+        "shortest_edge": 4096
+    },
+    "patch_size": 16,
+    "temporal_patch_size": 2,
+    "merge_size": 2,
+    "image_mean": [
+        0.5,
+        0.5,
+        0.5
+    ],
+    "image_std": [
+        0.5,
+        0.5,
+        0.5
+    ],
+    "processor_class": "Qwen3VLProcessor",
+    "video_processor_type": "Qwen3VLVideoProcessor"
+}

tokenizer/tokenizer_config.json

@@ -0,0 +1,246 @@
+{
+  "add_bos_token": false,
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151646": {
+      "content": "<|object_ref_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151647": {
+      "content": "<|object_ref_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151648": {
+      "content": "<|box_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151649": {
+      "content": "<|box_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151650": {
+      "content": "<|quad_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151651": {
+      "content": "<|quad_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151652": {
+      "content": "<|vision_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151653": {
+      "content": "<|vision_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151654": {
+      "content": "<|vision_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151655": {
+      "content": "<|image_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151656": {
+      "content": "<|video_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151657": {
+      "content": "<tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151658": {
+      "content": "</tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151659": {
+      "content": "<|fim_prefix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151660": {
+      "content": "<|fim_middle|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151661": {
+      "content": "<|fim_suffix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151662": {
+      "content": "<|fim_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151663": {
+      "content": "<|repo_name|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151664": {
+      "content": "<|file_sep|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151665": {
+      "content": "<tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151666": {
+      "content": "</tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151667": {
+      "content": "<think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151668": {
+      "content": "</think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>",
+    "<d>",
+    "</d>",
+    "<|cutoff|>",
+    "<|lyrics_start|>",
+    "<|lyrics_end|>",
+    "<|caption_start|>",
+    "<|caption_end|>"
+  ],
+  "bos_token": null,
+  "chat_template": "{%- if tools %}\n    {{- '<|im_start|>system\\n' }}\n    {%- if messages[0].role == 'system' %}\n        {%- if messages[0].content is string %}\n            {{- messages[0].content }}\n        {%- else %}\n            {%- for content in messages[0].content %}\n                {%- if 'text' in content %}\n                    {{- content.text }}\n                {%- endif %}\n            {%- endfor %}\n        {%- endif %}\n        {{- '\\n\\n' }}\n    {%- endif %}\n    {{- \"# Tools\\n\\nYou may call one or more functions to assist with the user query.\\n\\nYou are provided with function signatures within <tools></tools> XML tags:\\n<tools>\" }}\n    {%- for tool in tools %}\n        {{- \"\\n\" }}\n        {{- tool | tojson }}\n    {%- endfor %}\n    {{- \"\\n</tools>\\n\\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\\n<tool_call>\\n{\\\"name\\\": <function-name>, \\\"arguments\\\": <args-json-object>}\\n</tool_call><|im_end|>\\n\" }}\n{%- else %}\n    {%- if messages[0].role == 'system' %}\n        {{- '<|im_start|>system\\n' }}\n        {%- if messages[0].content is string %}\n            {{- messages[0].content }}\n        {%- else %}\n            {%- for content in messages[0].content %}\n                {%- if 'text' in content %}\n                    {{- content.text }}\n                {%- endif %}\n            {%- endfor %}\n        {%- endif %}\n        {{- '<|im_end|>\\n' }}\n    {%- endif %}\n{%- endif %}\n{%- set image_count = namespace(value=0) %}\n{%- set video_count = namespace(value=0) %}\n{%- for message in messages %}\n    {%- if message.role == \"user\" %}\n        {{- '<|im_start|>' + message.role + '\\n' }}\n        {%- if message.content is string %}\n            {{- message.content }}\n        {%- else %}\n            {%- for content in message.content %}\n                {%- if content.type == 'image' or 'image' in content or 'image_url' in content %}\n                    {%- set image_count.value = image_count.value + 1 %}\n                    {%- if add_vision_id %}Picture {{ image_count.value }}: {% endif -%}\n                    <|vision_start|><|image_pad|><|vision_end|>\n                {%- elif content.type == 'video' or 'video' in content %}\n                    {%- set video_count.value = video_count.value + 1 %}\n                    {%- if add_vision_id %}Video {{ video_count.value }}: {% endif -%}\n                    <|vision_start|><|video_pad|><|vision_end|>\n                {%- elif 'text' in content %}\n                    {{- content.text }}\n                {%- endif %}\n            {%- endfor %}\n        {%- endif %}\n        {{- '<|im_end|>\\n' }}\n    {%- elif message.role == \"assistant\" %}\n        {{- '<|im_start|>' + message.role + '\\n' }}\n        {%- if message.content is string %}\n            {{- message.content }}\n        {%- else %}\n            {%- for content_item in message.content %}\n                {%- if 'text' in content_item %}\n                    {{- content_item.text }}\n                {%- endif %}\n            {%- endfor %}\n        {%- endif %}\n        {%- if message.tool_calls %}\n            {%- for tool_call in message.tool_calls %}\n                {%- if (loop.first and message.content) or (not loop.first) %}\n                    {{- '\\n' }}\n                {%- endif %}\n                {%- if tool_call.function %}\n                    {%- set tool_call = tool_call.function %}\n                {%- endif %}\n                {{- '<tool_call>\\n{\"name\": \"' }}\n                {{- tool_call.name }}\n                {{- '\", \"arguments\": ' }}\n                {%- if tool_call.arguments is string %}\n                    {{- tool_call.arguments }}\n                {%- else %}\n                    {{- tool_call.arguments | tojson }}\n                {%- endif %}\n                {{- '}\\n</tool_call>' }}\n            {%- endfor %}\n        {%- endif %}\n        {{- '<|im_end|>\\n' }}\n    {%- elif message.role == \"tool\" %}\n        {%- if loop.first or (messages[loop.index0 - 1].role != \"tool\") %}\n            {{- '<|im_start|>user' }}\n        {%- endif %}\n        {{- '\\n<tool_response>\\n' }}\n        {%- if message.content is string %}\n            {{- message.content }}\n        {%- else %}\n            {%- for content in message.content %}\n                {%- if content.type == 'image' or 'image' in content or 'image_url' in content %}\n                    {%- set image_count.value = image_count.value + 1 %}\n                    {%- if add_vision_id %}Picture {{ image_count.value }}: {% endif -%}\n                    <|vision_start|><|image_pad|><|vision_end|>\n                {%- elif content.type == 'video' or 'video' in content %}\n                    {%- set video_count.value = video_count.value + 1 %}\n                    {%- if add_vision_id %}Video {{ video_count.value }}: {% endif -%}\n                    <|vision_start|><|video_pad|><|vision_end|>\n                {%- elif 'text' in content %}\n                    {{- content.text }}\n                {%- endif %}\n            {%- endfor %}\n        {%- endif %}\n        {{- '\\n</tool_response>' }}\n        {%- if loop.last or (messages[loop.index0 + 1].role != \"tool\") %}\n            {{- '<|im_end|>\\n' }}\n        {%- endif %}\n    {%- endif %}\n{%- endfor %}\n{%- if add_generation_prompt %}\n    {{- '<|im_start|>assistant\\n' }}\n{%- endif %}\n",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "model_max_length": 262144,
+  "pad_token": "<|endoftext|>",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}

transformer/config.json

@@ -0,0 +1,34 @@
+{
+  "_class_name": "MiniMaxH3Transformer3DModel",
+  "_diffusers_version": "0.36.0.dev0",
+  "attention_head_dim": 128,
+  "audio_in_channels": 32,
+  "ffn_dim": 14336,
+  "final_norm_eps": 1e-05,
+  "freq_dim": 256,
+  "hidden_size": 5376,
+  "in_channels": 24,
+  "norm_eps": 1e-05,
+  "num_attention_heads": 56,
+  "num_layers": 50,
+  "num_refiner_layers": 2,
+  "patch_size": [
+    1,
+    2,
+    2
+  ],
+  "pdd_steps": 32,
+  "qk_norm_eps": 1e-05,
+  "rope_freq_dim": 16,
+  "rope_theta": 10000.0,
+  "text_dim": 5120,
+  "time_embed_dim": 2688,
+  "time_embed_hidden_dim": 5376,
+  "quantization_config": {
+    "quant_method": "fp8",
+    "fmt": "e4m3",
+    "weight_scheme": "per_channel",
+    "activation_scheme": "dynamic_per_token",
+    "quantized_modules": "^transformer_blocks\\.\\d+\\.(attn\\.to_q|attn\\.to_k|attn\\.to_v|attn\\.to_out\\.0|ff\\.net\\.0\\.proj|ff\\.net\\.2)\\.weight$"
+  }
+}

transformer/diffusion_pytorch_model-00001-of-00014.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1397e48aceb4bb718206bc59d027d9c9beeb0c735da10aac4dcf148a92ceec20
+size 4410869144

transformer/diffusion_pytorch_model-00003-of-00014.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c17296377b9c47992030422506993226dbb6058734b750f13ad82c6105a68c3c
+size 3816798528

transformer/diffusion_pytorch_model-00011-of-00014.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:050d77a2079aa53c599860539e01c91e2d427e67cbc2133e1c488a49ce613811
+size 3778177456

transformer/diffusion_pytorch_model-00012-of-00014.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:89979cc4642291bd40373460ca6edc0e34f0fd6271ac987433243a04a1950491
+size 3855304928

transformer/diffusion_pytorch_model-00013-of-00014.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:50a47ea1aa7fcf585a7cc650fad12f4fd833cc648143fb666155b52a2cee6fb1
+size 3855312016

transformer/diffusion_pytorch_model-00014-of-00014.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9059e306265e4ddaa36ecde9c14f707d6ffd51f855ee8b2b628200c970eb6a57
+size 1368726728

transformer_ref/config.json

@@ -0,0 +1,34 @@
+{
+  "_class_name": "MiniMaxH3Transformer3DModel",
+  "_diffusers_version": "0.36.0.dev0",
+  "attention_head_dim": 128,
+  "audio_in_channels": 32,
+  "ffn_dim": 14336,
+  "final_norm_eps": 1e-05,
+  "freq_dim": 256,
+  "hidden_size": 5376,
+  "in_channels": 24,
+  "norm_eps": 1e-05,
+  "num_attention_heads": 56,
+  "num_layers": 50,
+  "num_refiner_layers": 2,
+  "patch_size": [
+    1,
+    2,
+    2
+  ],
+  "pdd_steps": 32,
+  "qk_norm_eps": 1e-05,
+  "rope_freq_dim": 16,
+  "rope_theta": 10000.0,
+  "text_dim": 5120,
+  "time_embed_dim": 2688,
+  "time_embed_hidden_dim": 5376,
+  "quantization_config": {
+    "quant_method": "fp8",
+    "fmt": "e4m3",
+    "weight_scheme": "per_channel",
+    "activation_scheme": "dynamic_per_token",
+    "quantized_modules": "^transformer_blocks\\.\\d+\\.(attn\\.to_q|attn\\.to_k|attn\\.to_v|attn\\.to_out\\.0|ff\\.net\\.0\\.proj|ff\\.net\\.2)\\.weight$"
+  }
+}

transformer_ref/diffusion_pytorch_model-00001-of-00014.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1397e48aceb4bb718206bc59d027d9c9beeb0c735da10aac4dcf148a92ceec20
+size 4410869144

transformer_ref/diffusion_pytorch_model-00002-of-00014.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8c3a86c313e3eefc6fb69134cefa22c972fea75e04f11839396c1a7fd259d8db
+size 3855312080

transformer_ref/diffusion_pytorch_model-00003-of-00014.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c17296377b9c47992030422506993226dbb6058734b750f13ad82c6105a68c3c
+size 3816798528

transformer_ref/diffusion_pytorch_model-00004-of-00014.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dc18e08f4fd564b61eda4d2783f185cdf8e7c7e34b1238e2bb4d3fc5d892e09b
+size 3412207416

transformer_ref/diffusion_pytorch_model-00009-of-00014.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5273bd68e3b280d4359d5beda17870936496419227c2bc12e56677b82ab9b706
+size 3335136960

transformer_ref/diffusion_pytorch_model-00013-of-00014.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:50a47ea1aa7fcf585a7cc650fad12f4fd833cc648143fb666155b52a2cee6fb1
+size 3855312016

vae/config.json

@@ -0,0 +1,97 @@
+{
+  "_class_name": "AutoencoderKLMiniMaxH3",
+  "_diffusers_version": "0.36.0.dev0",
+  "in_channels": 3,
+  "out_channels": 3,
+  "latent_channels": 24,
+  "block_out_channels": [
+    128,
+    256,
+    256,
+    512,
+    512,
+    1024
+  ],
+  "layers_per_block": 2,
+  "spatial_downsample_factors": [
+    2,
+    2,
+    2,
+    2,
+    1,
+    1
+  ],
+  "temporal_downsample_factors": [
+    1,
+    2,
+    2,
+    1,
+    1,
+    1
+  ],
+  "norm_num_groups": 32,
+  "norm_eps": 1e-06,
+  "spatial_padding_mode": "reflect",
+  "decoder_num_layers": 26,
+  "decoder_num_attention_heads": 32,
+  "decoder_attention_head_dim": 64,
+  "decoder_num_register_tokens": 4,
+  "decoder_ffn_mult": 4,
+  "decoder_rope_theta": 100.0,
+  "decoder_rope_dim_ratio": 0.75,
+  "decoder_norm_eps": 1e-05,
+  "clip_length": 17,
+  "token_drop": 3,
+  "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
+  ]
+}

vae/diffusion_pytorch_model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:73fc5a61caeba438633eb8f8efd3a57f4915adfa46731a1c991312c724a3b5cc
+size 4225603880