Hugging Face's logo

corechan
/
MiniMax-H3_NF4

Text Generation
Diffusers
Safetensors
text-to-video
video-generation
quantized
nf4
bitsandbytes
Model card Files
xet
 Community
corechan commited on
Commit
e477a62
·
verified ·
1 Parent(s): 055256f

quantized text_encoder (int8) from memory

Browse files
Files changed (4) hide show
  1. quantized/text_encoder-int8/.complete +1 -0
  2. quantized/text_encoder-int8/config.json +94 -0
  3. quantized/text_encoder-int8/generation_config.json +7 -0
  4. quantized/text_encoder-int8/pytorch_model.bin +3 -0
quantized/text_encoder-int8/.complete ADDED
@@ -0,0 +1 @@
 
 
1
+ {"component": "text_encoder", "quant": "int8", "ts": 1789204831.039434, "safetensors": false, "torchao": "0.15.0"}
quantized/text_encoder-int8/config.json ADDED
@@ -0,0 +1,94 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Qwen3VLForConditionalGeneration"
4
+ ],
5
+ "dtype": "bfloat16",
6
+ "image_token_id": 151655,
7
+ "model_type": "qwen3_vl",
8
+ "quantization_config": {
9
+ "include_input_output_embeddings": false,
10
+ "modules_to_not_convert": [
11
+ "model.visual",
12
+ "model.language_model.embed_tokens",
13
+ "model.language_model.norm",
14
+ "lm_head"
15
+ ],
16
+ "quant_method": "torchao",
17
+ "quant_type": {
18
+ "default": {
19
+ "_data": {
20
+ "granularity": {
21
+ "_data": {
22
+ "dim": -1
23
+ },
24
+ "_type": "PerRow",
25
+ "_version": 1
26
+ },
27
+ "group_size": null,
28
+ "set_inductor_config": true
29
+ },
30
+ "_type": "Int8WeightOnlyConfig",
31
+ "_version": 2
32
+ }
33
+ },
34
+ "untie_embedding_weights": false
35
+ },
36
+ "text_config": {
37
+ "attention_bias": false,
38
+ "attention_dropout": 0.0,
39
+ "bos_token_id": 151643,
40
+ "dtype": "bfloat16",
41
+ "eos_token_id": 151645,
42
+ "head_dim": 128,
43
+ "hidden_act": "silu",
44
+ "hidden_size": 5120,
45
+ "initializer_range": 0.02,
46
+ "intermediate_size": 25600,
47
+ "max_position_embeddings": 262144,
48
+ "model_type": "qwen3_vl_text",
49
+ "num_attention_heads": 64,
50
+ "num_hidden_layers": 51,
51
+ "num_key_value_heads": 8,
52
+ "pad_token_id": null,
53
+ "rms_norm_eps": 1e-06,
54
+ "rope_parameters": {
55
+ "mrope_interleaved": true,
56
+ "mrope_section": [
57
+ 24,
58
+ 20,
59
+ 20
60
+ ],
61
+ "rope_theta": 5000000,
62
+ "rope_type": "default"
63
+ },
64
+ "tie_word_embeddings": true,
65
+ "use_cache": true,
66
+ "vocab_size": 151936
67
+ },
68
+ "tie_word_embeddings": true,
69
+ "transformers_version": "5.16.1",
70
+ "video_token_id": 151656,
71
+ "vision_config": {
72
+ "deepstack_visual_indexes": [
73
+ 8,
74
+ 16,
75
+ 24
76
+ ],
77
+ "depth": 27,
78
+ "dtype": "bfloat16",
79
+ "hidden_act": "gelu_pytorch_tanh",
80
+ "hidden_size": 1152,
81
+ "in_channels": 3,
82
+ "initializer_range": 0.02,
83
+ "intermediate_size": 4304,
84
+ "model_type": "qwen3_vl_vision",
85
+ "num_heads": 16,
86
+ "num_position_embeddings": 2304,
87
+ "out_hidden_size": 5120,
88
+ "patch_size": 16,
89
+ "spatial_merge_size": 2,
90
+ "temporal_patch_size": 2
91
+ },
92
+ "vision_end_token_id": 151653,
93
+ "vision_start_token_id": 151652
94
+ }
quantized/text_encoder-int8/generation_config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 151643,
4
+ "eos_token_id": 151645,
5
+ "transformers_version": "5.16.1",
6
+ "use_cache": true
7
+ }
quantized/text_encoder-int8/pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1131b16dac6c4e04fc87144081eb812bd6c77eace06c6413bf303a05885068e7
3
+ size 27622282854