Hugging Face's logo

MiniMaxAI
/
MiniMax-H3

Image-Text-to-Video
Diffusers
Safetensors
MiniMaxH3ModularPipeline
text-to-video
image-to-video
video-to-video
text-to-audio-video
image-to-audio-video
image-text-to-audio-video
video-to-audio-video
audio-to-audio-video
audio-video-generation
multimodal
synchronized-audio-video
reference-to-audio-video
Model card Files
xet
 Community
71
yuanhe commited on
Commit
48d93ed
·
1 Parent(s): 5d9b308

Add repository model index for HF downloads

Browse files
Files changed (2) hide show
  1. README.md +4 -2
  2. model_index.json +139 -0
README.md CHANGED
@@ -194,12 +194,14 @@ Each checkpoint is distributed as a self\-contained Hugging Face\-style reposito
194
 
195
  Download the model. The repository hosts the original checkpoint (`FL2VA/`, `Ref2VA/`) and the diffusers format side by side, so scope the download to what your framework needs:
196
 
 
 
197
  ```bash
198
  # Original checkpoint, both task families (SGLang, vLLM):
199
- hf download MiniMaxAI/MiniMax-H3 --include "FL2VA/*" "Ref2VA/*" --local-dir MiniMax-H3
200
 
201
  # Or a single task family:
202
- hf download MiniMaxAI/MiniMax-H3 --include "FL2VA/*" --local-dir MiniMax-H3
203
  ```
204
 
205
  diffusers users do not need a manual download: `ModularPipeline.from_pretrained("MiniMaxAI/MiniMax-H3")` fetches exactly the components it needs. See the [diffusers documentation](https://github.com/huggingface/diffusers/blob/minimax-h3/docs/source/en/api/pipelines/minimax_h3.md) for loading recipes.
 
194
 
195
  Download the model. The repository hosts the original checkpoint (`FL2VA/`, `Ref2VA/`) and the diffusers format side by side, so scope the download to what your framework needs:
196
 
197
+ `model_index.json` is the repository-level modular index. The task-family-specific diffusers indexes remain under `FL2VA/model_index.json` and `Ref2VA/model_index.json`.
198
+
199
  ```bash
200
  # Original checkpoint, both task families (SGLang, vLLM):
201
+ hf download MiniMaxAI/MiniMax-H3 --include "model_index.json" "modular_model_index.json" "FL2VA/*" "Ref2VA/*" --local-dir MiniMax-H3
202
 
203
  # Or a single task family:
204
+ hf download MiniMaxAI/MiniMax-H3 --include "model_index.json" "modular_model_index.json" "FL2VA/*" --local-dir MiniMax-H3
205
  ```
206
 
207
  diffusers users do not need a manual download: `ModularPipeline.from_pretrained("MiniMaxAI/MiniMax-H3")` fetches exactly the components it needs. See the [diffusers documentation](https://github.com/huggingface/diffusers/blob/minimax-h3/docs/source/en/api/pipelines/minimax_h3.md) for loading recipes.
model_index.json ADDED
@@ -0,0 +1,139 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_class_name": "MiniMaxH3ModularPipeline",
3
+ "_diffusers_version": "0.36.0.dev0",
4
+ "_blocks_class_name": "MiniMaxH3Blocks",
5
+ "_minimax_h3": {
6
+ "schema_version": 1,
7
+ "index_scope": "repository",
8
+ "task_family_indexes": {
9
+ "fl2va": "FL2VA/model_index.json",
10
+ "ref2va": "Ref2VA/model_index.json"
11
+ }
12
+ },
13
+ "text_encoder": [
14
+ "transformers",
15
+ "Qwen3VLForConditionalGeneration",
16
+ {
17
+ "type_hint": [
18
+ "transformers",
19
+ "Qwen3VLForConditionalGeneration"
20
+ ],
21
+ "pretrained_model_name_or_path": "MiniMaxAI/MiniMax-H3",
22
+ "subfolder": "text_encoder",
23
+ "variant": null,
24
+ "revision": null
25
+ }
26
+ ],
27
+ "tokenizer": [
28
+ "transformers",
29
+ "Qwen2TokenizerFast",
30
+ {
31
+ "type_hint": [
32
+ "transformers",
33
+ "Qwen2TokenizerFast"
34
+ ],
35
+ "pretrained_model_name_or_path": "MiniMaxAI/MiniMax-H3",
36
+ "subfolder": "tokenizer",
37
+ "variant": null,
38
+ "revision": null
39
+ }
40
+ ],
41
+ "processor": [
42
+ "transformers",
43
+ "Qwen3VLProcessor",
44
+ {
45
+ "type_hint": [
46
+ "transformers",
47
+ "Qwen3VLProcessor"
48
+ ],
49
+ "pretrained_model_name_or_path": "MiniMaxAI/MiniMax-H3",
50
+ "subfolder": "processor",
51
+ "variant": null,
52
+ "revision": null
53
+ }
54
+ ],
55
+ "vae": [
56
+ "diffusers",
57
+ "AutoencoderKLMiniMaxH3",
58
+ {
59
+ "type_hint": [
60
+ "diffusers",
61
+ "AutoencoderKLMiniMaxH3"
62
+ ],
63
+ "pretrained_model_name_or_path": "MiniMaxAI/MiniMax-H3",
64
+ "subfolder": "vae",
65
+ "variant": null,
66
+ "revision": null
67
+ }
68
+ ],
69
+ "audio_vae": [
70
+ "diffusers",
71
+ "AutoencoderKLMiniMaxH3Audio",
72
+ {
73
+ "type_hint": [
74
+ "diffusers",
75
+ "AutoencoderKLMiniMaxH3Audio"
76
+ ],
77
+ "pretrained_model_name_or_path": "MiniMaxAI/MiniMax-H3",
78
+ "subfolder": "audio_vae",
79
+ "variant": null,
80
+ "revision": null
81
+ }
82
+ ],
83
+ "transformer": [
84
+ "diffusers",
85
+ "MiniMaxH3Transformer3DModel",
86
+ {
87
+ "type_hint": [
88
+ "diffusers",
89
+ "MiniMaxH3Transformer3DModel"
90
+ ],
91
+ "pretrained_model_name_or_path": "MiniMaxAI/MiniMax-H3",
92
+ "subfolder": "transformer",
93
+ "variant": null,
94
+ "revision": null
95
+ }
96
+ ],
97
+ "transformer_ref": [
98
+ "diffusers",
99
+ "MiniMaxH3Transformer3DModel",
100
+ {
101
+ "type_hint": [
102
+ "diffusers",
103
+ "MiniMaxH3Transformer3DModel"
104
+ ],
105
+ "pretrained_model_name_or_path": "MiniMaxAI/MiniMax-H3",
106
+ "subfolder": "transformer_ref",
107
+ "variant": null,
108
+ "revision": null
109
+ }
110
+ ],
111
+ "scheduler": [
112
+ "diffusers",
113
+ "MiniMaxH3Scheduler",
114
+ {
115
+ "type_hint": [
116
+ "diffusers",
117
+ "MiniMaxH3Scheduler"
118
+ ],
119
+ "pretrained_model_name_or_path": "MiniMaxAI/MiniMax-H3",
120
+ "subfolder": "scheduler",
121
+ "variant": null,
122
+ "revision": null
123
+ }
124
+ ],
125
+ "audio_scheduler": [
126
+ "diffusers",
127
+ "MiniMaxH3Scheduler",
128
+ {
129
+ "type_hint": [
130
+ "diffusers",
131
+ "MiniMaxH3Scheduler"
132
+ ],
133
+ "pretrained_model_name_or_path": "MiniMaxAI/MiniMax-H3",
134
+ "subfolder": "audio_scheduler",
135
+ "variant": null,
136
+ "revision": null
137
+ }
138
+ ]
139
+ }