wlsaidhi commited on
Commit
0986355
·
verified ·
1 Parent(s): a8fa1e1

[docs]: align FastH3 Preview v1 naming

Browse files

Update the model card and package self-references after the server-side repository rename.

README.md CHANGED
@@ -16,16 +16,18 @@ tags:
16
  - minimax-h3
17
  - fastvideo
18
  - fasth3
 
19
  ---
20
 
21
- # FastVideo-FastH3-4-step-v1.1
22
 
23
- Private FastVideo evaluation snapshot of the `dmd2_sp2_fsdp64_v11_dataonly_mixed_vsa64` MiniMax-H3
24
- student at training step 1300. It generates synchronized video and audio
25
- with four transformer forwards.
 
26
 
27
- This repository is intentionally private. It is an internal checkpoint for
28
- evaluation and UI integration, not a public release recommendation.
29
 
30
  ## Checkpoint identity
31
 
@@ -60,7 +62,7 @@ four-point scheduler grid.
60
  export FASTVIDEO_DMD_DENOISING_STEPS=999,749,500,250
61
 
62
  python examples/inference/basic/basic_fasth3.py \
63
- --model-path FastVideo/FastVideo-FastH3-4-step-v1.1 \
64
  --prompt '<MiniMax-H3 T2AV prompt>' \
65
  --steps 5 \
66
  --num-gpus 4 \
@@ -70,22 +72,24 @@ python examples/inference/basic/basic_fasth3.py \
70
  ```
71
 
72
  The five scheduler grid points execute the four trained jumps. Use guidance
73
- scale 1.0 and VSA-H3 at 90% sparsity with 64-token tiles. Compilation is an optional runtime
74
- optimization; first validate output parity on the pinned FastVideo commit.
 
75
 
76
  ## Training summary
77
 
78
- - Method: data-only DMD2 on synthetic base-H3 video latents. The student is trained from forward-noised target latents; no data-free carried rollout is mixed into this run.
 
 
79
  - Student grid: `[999, 749, 500, 250]`.
80
  - Student attention: VSA-H3 at 90% sparsity with 64-token tiles.
81
  - Global batch size: 64.
82
  - Student and critic learning rate: `2e-06` and
83
  `2e-06`.
84
  - Precision: fp32 training state and bf16 inference export.
85
- - Training topology: 64 GB200 GPUs, SP=2,
86
  HSDP replicate=1 / shard=64.
87
- - Strict continuous score clock: `True`;
88
- shift `2.4`, warp max `0.999`.
89
 
90
  ## Limitations
91
 
 
16
  - minimax-h3
17
  - fastvideo
18
  - fasth3
19
+ - preview
20
  ---
21
 
22
+ # FastVideo-FastH3-4-step-Preview-v1-VSA-Synthetic-Step1300
23
 
24
+ FastH3 4-Step Preview v1 checkpoint from FastVideo. This variant is the
25
+ `dmd2_sp2_fsdp64_v11_dataonly_mixed_vsa64` synthetic-data DMD2 MiniMax-H3
26
+ student at training step 1300. It generates synchronized video and audio with
27
+ four transformer forwards and uses VSA-H3 at 90% sparsity.
28
 
29
+ See the complete [FastH3 4-Step Preview v1
30
+ collection](https://huggingface.co/collections/FastVideo/fastvideo-fasth3).
31
 
32
  ## Checkpoint identity
33
 
 
62
  export FASTVIDEO_DMD_DENOISING_STEPS=999,749,500,250
63
 
64
  python examples/inference/basic/basic_fasth3.py \
65
+ --model-path FastVideo/FastVideo-FastH3-4-step-Preview-v1-VSA-Synthetic-Step1300 \
66
  --prompt '<MiniMax-H3 T2AV prompt>' \
67
  --steps 5 \
68
  --num-gpus 4 \
 
72
  ```
73
 
74
  The five scheduler grid points execute the four trained jumps. Use guidance
75
+ scale 1.0 and VSA-H3 at 90% sparsity with 64-token tiles. Compilation is an
76
+ optional runtime optimization; first validate output parity on the pinned
77
+ FastVideo commit.
78
 
79
  ## Training summary
80
 
81
+ - Method: data-only DMD2 on synthetic Base-H3 video latents. The student is
82
+ trained from forward-noised target latents; no data-free carried rollout is
83
+ mixed into this run.
84
  - Student grid: `[999, 749, 500, 250]`.
85
  - Student attention: VSA-H3 at 90% sparsity with 64-token tiles.
86
  - Global batch size: 64.
87
  - Student and critic learning rate: `2e-06` and
88
  `2e-06`.
89
  - Precision: fp32 training state and bf16 inference export.
90
+ - Training topology: 64 B200 GPUs, SP=2,
91
  HSDP replicate=1 / shard=64.
92
+ - Continuous score-clock shift: `2.4`; warp max: `0.999`.
 
93
 
94
  ## Limitations
95
 
fastvideo_inference.json CHANGED
@@ -11,7 +11,7 @@
11
  ],
12
  "fastvideo_commit": "6ac300b5a7c472e3a139f1cb3ff7a03b58f3e26e",
13
  "guidance_scale": 1.0,
14
- "model_id": "FastVideo/FastVideo-FastH3-4-step-v1.1",
15
  "num_gpus": 4,
16
  "num_inference_steps": 5,
17
  "schema_version": "fasth3-inference-contract-v1",
 
11
  ],
12
  "fastvideo_commit": "6ac300b5a7c472e3a139f1cb3ff7a03b58f3e26e",
13
  "guidance_scale": 1.0,
14
+ "model_id": "FastVideo/FastVideo-FastH3-4-step-Preview-v1-VSA-Synthetic-Step1300",
15
  "num_gpus": 4,
16
  "num_inference_steps": 5,
17
  "schema_version": "fasth3-inference-contract-v1",
modular_model_index.json CHANGED
@@ -6,7 +6,7 @@
6
  "diffusers",
7
  "MiniMaxH3Scheduler",
8
  {
9
- "pretrained_model_name_or_path": "FastVideo/FastVideo-FastH3-4-step-v1.1",
10
  "revision": null,
11
  "subfolder": "audio_scheduler",
12
  "type_hint": [
@@ -20,7 +20,7 @@
20
  "diffusers",
21
  "AutoencoderKLMiniMaxH3Audio",
22
  {
23
- "pretrained_model_name_or_path": "FastVideo/FastVideo-FastH3-4-step-v1.1",
24
  "revision": null,
25
  "subfolder": "audio_vae",
26
  "type_hint": [
@@ -34,7 +34,7 @@
34
  "transformers",
35
  "Qwen3VLProcessor",
36
  {
37
- "pretrained_model_name_or_path": "FastVideo/FastVideo-FastH3-4-step-v1.1",
38
  "revision": null,
39
  "subfolder": "processor",
40
  "type_hint": [
@@ -48,7 +48,7 @@
48
  "diffusers",
49
  "MiniMaxH3Scheduler",
50
  {
51
- "pretrained_model_name_or_path": "FastVideo/FastVideo-FastH3-4-step-v1.1",
52
  "revision": null,
53
  "subfolder": "scheduler",
54
  "type_hint": [
@@ -62,7 +62,7 @@
62
  "transformers",
63
  "Qwen3VLForConditionalGeneration",
64
  {
65
- "pretrained_model_name_or_path": "FastVideo/FastVideo-FastH3-4-step-v1.1",
66
  "revision": null,
67
  "subfolder": "text_encoder",
68
  "type_hint": [
@@ -76,7 +76,7 @@
76
  "transformers",
77
  "Qwen2TokenizerFast",
78
  {
79
- "pretrained_model_name_or_path": "FastVideo/FastVideo-FastH3-4-step-v1.1",
80
  "revision": null,
81
  "subfolder": "tokenizer",
82
  "type_hint": [
@@ -90,7 +90,7 @@
90
  "diffusers",
91
  "MiniMaxH3Transformer3DModel",
92
  {
93
- "pretrained_model_name_or_path": "FastVideo/FastVideo-FastH3-4-step-v1.1",
94
  "revision": null,
95
  "subfolder": "transformer",
96
  "type_hint": [
@@ -118,7 +118,7 @@
118
  "diffusers",
119
  "AutoencoderKLMiniMaxH3",
120
  {
121
- "pretrained_model_name_or_path": "FastVideo/FastVideo-FastH3-4-step-v1.1",
122
  "revision": null,
123
  "subfolder": "vae",
124
  "type_hint": [
 
6
  "diffusers",
7
  "MiniMaxH3Scheduler",
8
  {
9
+ "pretrained_model_name_or_path": "FastVideo/FastVideo-FastH3-4-step-Preview-v1-VSA-Synthetic-Step1300",
10
  "revision": null,
11
  "subfolder": "audio_scheduler",
12
  "type_hint": [
 
20
  "diffusers",
21
  "AutoencoderKLMiniMaxH3Audio",
22
  {
23
+ "pretrained_model_name_or_path": "FastVideo/FastVideo-FastH3-4-step-Preview-v1-VSA-Synthetic-Step1300",
24
  "revision": null,
25
  "subfolder": "audio_vae",
26
  "type_hint": [
 
34
  "transformers",
35
  "Qwen3VLProcessor",
36
  {
37
+ "pretrained_model_name_or_path": "FastVideo/FastVideo-FastH3-4-step-Preview-v1-VSA-Synthetic-Step1300",
38
  "revision": null,
39
  "subfolder": "processor",
40
  "type_hint": [
 
48
  "diffusers",
49
  "MiniMaxH3Scheduler",
50
  {
51
+ "pretrained_model_name_or_path": "FastVideo/FastVideo-FastH3-4-step-Preview-v1-VSA-Synthetic-Step1300",
52
  "revision": null,
53
  "subfolder": "scheduler",
54
  "type_hint": [
 
62
  "transformers",
63
  "Qwen3VLForConditionalGeneration",
64
  {
65
+ "pretrained_model_name_or_path": "FastVideo/FastVideo-FastH3-4-step-Preview-v1-VSA-Synthetic-Step1300",
66
  "revision": null,
67
  "subfolder": "text_encoder",
68
  "type_hint": [
 
76
  "transformers",
77
  "Qwen2TokenizerFast",
78
  {
79
+ "pretrained_model_name_or_path": "FastVideo/FastVideo-FastH3-4-step-Preview-v1-VSA-Synthetic-Step1300",
80
  "revision": null,
81
  "subfolder": "tokenizer",
82
  "type_hint": [
 
90
  "diffusers",
91
  "MiniMaxH3Transformer3DModel",
92
  {
93
+ "pretrained_model_name_or_path": "FastVideo/FastVideo-FastH3-4-step-Preview-v1-VSA-Synthetic-Step1300",
94
  "revision": null,
95
  "subfolder": "transformer",
96
  "type_hint": [
 
118
  "diffusers",
119
  "AutoencoderKLMiniMaxH3",
120
  {
121
+ "pretrained_model_name_or_path": "FastVideo/FastVideo-FastH3-4-step-Preview-v1-VSA-Synthetic-Step1300",
122
  "revision": null,
123
  "subfolder": "vae",
124
  "type_hint": [
provenance.json CHANGED
@@ -13,7 +13,7 @@
13
  ],
14
  "fastvideo_commit": "6ac300b5a7c472e3a139f1cb3ff7a03b58f3e26e",
15
  "lineage_parent_run_id": "20260823T191211Z-v11-dataonly-sp2-initial",
16
- "model_id": "FastVideo/FastVideo-FastH3-4-step-v1.1",
17
  "schema_version": "fasth3-checkpoint-provenance-v1",
18
  "source_config_sha256": "b1e0bd319ba7dddb8ff0a9db9bcdb3182b9d645c43fe3fcce60d678a092ae8c5",
19
  "student_dtype": "bfloat16",
 
13
  ],
14
  "fastvideo_commit": "6ac300b5a7c472e3a139f1cb3ff7a03b58f3e26e",
15
  "lineage_parent_run_id": "20260823T191211Z-v11-dataonly-sp2-initial",
16
+ "model_id": "FastVideo/FastVideo-FastH3-4-step-Preview-v1-VSA-Synthetic-Step1300",
17
  "schema_version": "fasth3-checkpoint-provenance-v1",
18
  "source_config_sha256": "b1e0bd319ba7dddb8ff0a9db9bcdb3182b9d645c43fe3fcce60d678a092ae8c5",
19
  "student_dtype": "bfloat16",