Hugging Face's logo

FractalSurfer
/
gemma-3-12b-it-norm-preserved-biprojected-abliterated-mlx-4Bit

Image-Text-to-Text
Transformers
Safetensors
MLX
gemma3
any-to-any
conversational
text-generation-inference
4-bit precision
Model card Files
xet
Community
FractalSurfer commited on
Commit
ec567d8
·
verified ·
1 Parent(s): eecf3f9

Upload config.json with huggingface_hub

Browse files
Files changed (1) hide show
  1. config.json +107 -0
config.json ADDED
@@ -0,0 +1,107 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Gemma3ForConditionalGeneration"
4
+ ],
5
+ "boi_token_index": 255999,
6
+ "dtype": "bfloat16",
7
+ "eoi_token_index": 256000,
8
+ "eos_token_id": [
9
+ 1,
10
+ 106
11
+ ],
12
+ "image_token_index": 262144,
13
+ "initializer_range": 0.02,
14
+ "mm_tokens_per_image": 256,
15
+ "model_type": "gemma3",
16
+ "quantization": {
17
+ "group_size": 64,
18
+ "bits": 4,
19
+ "mode": "affine"
20
+ },
21
+ "quantization_config": {
22
+ "group_size": 64,
23
+ "bits": 4,
24
+ "mode": "affine"
25
+ },
26
+ "text_config": {
27
+ "_sliding_window_pattern": 6,
28
+ "attention_bias": false,
29
+ "attention_dropout": 0.0,
30
+ "attn_logit_softcapping": null,
31
+ "dtype": "bfloat16",
32
+ "final_logit_softcapping": null,
33
+ "head_dim": 256,
34
+ "hidden_activation": "gelu_pytorch_tanh",
35
+ "hidden_size": 3840,
36
+ "initializer_range": 0.02,
37
+ "intermediate_size": 15360,
38
+ "layer_types": [
39
+ "sliding_attention",
40
+ "sliding_attention",
41
+ "sliding_attention",
42
+ "sliding_attention",
43
+ "sliding_attention",
44
+ "full_attention",
45
+ "sliding_attention",
46
+ "sliding_attention",
47
+ "sliding_attention",
48
+ "sliding_attention",
49
+ "sliding_attention",
50
+ "full_attention",
51
+ "sliding_attention",
52
+ "sliding_attention",
53
+ "sliding_attention",
54
+ "sliding_attention",
55
+ "sliding_attention",
56
+ "full_attention",
57
+ "sliding_attention",
58
+ "sliding_attention",
59
+ "sliding_attention",
60
+ "sliding_attention",
61
+ "sliding_attention",
62
+ "full_attention",
63
+ "sliding_attention",
64
+ "sliding_attention",
65
+ "sliding_attention",
66
+ "sliding_attention",
67
+ "sliding_attention",
68
+ "full_attention",
69
+ "sliding_attention",
70
+ "sliding_attention",
71
+ "sliding_attention",
72
+ "sliding_attention",
73
+ "sliding_attention",
74
+ "full_attention",
75
+ "sliding_attention",
76
+ "sliding_attention",
77
+ "sliding_attention",
78
+ "sliding_attention",
79
+ "sliding_attention",
80
+ "full_attention",
81
+ "sliding_attention",
82
+ "sliding_attention",
83
+ "sliding_attention",
84
+ "sliding_attention",
85
+ "sliding_attention",
86
+ "full_attention"
87
+ ],
88
+ "max_position_embeddings": 131072,
89
+ "model_type": "gemma3_text",
90
+ "num_attention_heads": 16,
91
+ "num_hidden_layers": 48,
92
+ "num_key_value_heads": 8,
93
+ "query_pre_attn_scalar": 256,
94
+ "rms_norm_eps": 1e-06,
95
+ "rope_local_base_freq": 10000.0,
96
+ "rope_scaling": {
97
+ "factor": 8.0,
98
+ "rope_type": "linear"
99
+ },
100
+ "rope_theta": 1000000.0,
101
+ "sliding_window": 1024,
102
+ "use_bidirectional_attention": false,
103
+ "use_cache": true,
104
+ "vocab_size": 262208
105
+ },
106
+ "transformers_version": "4.57.1"
107
+ }