Hugging Face's logo

unsloth
/
gemma-3-12b-pt-unsloth-bnb-4bit

Image-Text-to-Text
Transformers
Safetensors
English
gemma3
unsloth
gemma
google
text-generation-inference
4-bit precision
bitsandbytes
Model card Files
xet
Community
danielhanchen commited on
Commit
f40535d
·
verified ·
1 Parent(s): c8f85c4

Add files using upload-large-folder tool

Browse files
Files changed (6) hide show
  1. config.json +53 -57
  2. generation_config.json +8 -3
  3. model-00001-of-00003.safetensors +2 -2
  4. model-00002-of-00003.safetensors +2 -2
  5. model-00003-of-00003.safetensors +2 -2
  6. model.safetensors.index.json +0 -0
config.json CHANGED
@@ -25,85 +25,81 @@
25
  "multi_modal_projector",
26
  "merger",
27
  "modality_projection",
28
- "lm_head",
29
- "multi_modal_projector",
30
- "merger",
31
- "modality_projection",
32
- "language_model.model.layers.14.mlp",
33
- "language_model.model.layers.15.self_attn",
 
 
 
34
  "language_model.model.layers.2.mlp",
35
- "language_model.model.layers.18.self_attn",
36
- "language_model.model.layers.15.mlp",
37
- "language_model.model.layers.20.self_attn",
 
 
 
38
  "language_model.model.layers.11.mlp",
 
39
  "language_model.model.layers.4.mlp",
40
- "language_model.model.layers.0.mlp",
41
- "language_model.model.layers.17.self_attn",
42
- "language_model.model.layers.9.mlp",
43
- "language_model.model.layers.10.mlp",
44
- "language_model.model.layers.2.self_attn",
45
- "language_model.model.layers.20.mlp",
46
- "language_model.model.layers.16.mlp",
47
- "vision_tower.vision_model.encoder.layers.22.self_attn",
48
- "vision_tower.vision_model.encoder.layers.21.self_attn",
49
- "language_model.model.layers.8.mlp",
50
- "language_model.model.layers.1.mlp",
51
- "vision_tower.vision_model.encoder.layers.24.self_attn",
52
- "vision_tower.vision_model.encoder.layers.19.self_attn",
53
  "language_model.model.layers.7.mlp",
54
- "vision_tower.vision_model.encoder.layers.17.mlp",
55
- "vision_tower.vision_model.encoder.layers.15.mlp",
56
- "vision_tower.vision_model.encoder.layers.17.self_attn",
57
- "language_model.model.layers.6.mlp",
58
  "language_model.model.layers.3.mlp",
59
- "vision_tower.vision_model.encoder.layers.12.mlp",
60
- "vision_tower.vision_model.encoder.layers.26.self_attn",
61
- "language_model.model.layers.5.mlp",
62
- "vision_tower.vision_model.encoder.layers.24.mlp",
63
  "vision_tower.vision_model.encoder.layers.23.self_attn",
64
- "vision_tower.vision_model.encoder.layers.20.mlp",
65
- "vision_tower.vision_model.encoder.layers.25.self_attn",
66
- "vision_tower.vision_model.encoder.layers.20.self_attn",
67
  "vision_tower.vision_model.encoder.layers.19.mlp",
68
- "vision_tower.vision_model.encoder.layers.14.mlp",
 
69
  "vision_tower.vision_model.encoder.layers.18.self_attn",
 
 
 
 
70
  "vision_tower.vision_model.encoder.layers.16.self_attn",
71
- "vision_tower.vision_model.encoder.layers.10.mlp",
72
- "vision_tower.vision_model.encoder.layers.14.self_attn",
73
  "vision_tower.vision_model.encoder.layers.15.self_attn",
74
- "vision_tower.vision_model.encoder.layers.22.mlp",
75
- "vision_tower.vision_model.encoder.layers.13.self_attn",
76
  "vision_tower.vision_model.encoder.layers.13.mlp",
77
- "vision_tower.vision_model.encoder.layers.25.mlp",
 
 
78
  "vision_tower.vision_model.encoder.layers.23.mlp",
79
- "vision_tower.vision_model.encoder.layers.11.mlp",
80
- "vision_tower.vision_model.encoder.layers.21.mlp",
81
  "vision_tower.vision_model.encoder.layers.16.mlp",
 
 
82
  "vision_tower.vision_model.encoder.layers.9.self_attn",
 
 
 
83
  "vision_tower.vision_model.encoder.layers.12.self_attn",
84
- "vision_tower.vision_model.encoder.layers.18.mlp",
85
- "vision_tower.vision_model.encoder.layers.4.mlp",
86
- "vision_tower.vision_model.encoder.layers.7.mlp",
87
  "vision_tower.vision_model.encoder.layers.9.mlp",
88
- "vision_tower.vision_model.encoder.layers.11.self_attn",
89
- "vision_tower.vision_model.encoder.layers.2.mlp",
90
- "vision_tower.vision_model.encoder.layers.5.self_attn",
91
- "vision_tower.vision_model.encoder.layers.1.mlp",
92
- "vision_tower.vision_model.encoder.layers.10.self_attn",
93
- "vision_tower.vision_model.encoder.layers.8.mlp",
94
  "vision_tower.vision_model.encoder.layers.4.self_attn",
 
 
95
  "vision_tower.vision_model.encoder.layers.5.mlp",
 
 
96
  "vision_tower.vision_model.encoder.layers.7.self_attn",
97
- "vision_tower.vision_model.encoder.layers.6.mlp",
98
- "vision_tower.vision_model.encoder.layers.3.self_attn",
99
- "vision_tower.vision_model.encoder.layers.3.mlp",
100
  "vision_tower.vision_model.encoder.layers.1.self_attn",
101
- "vision_tower.vision_model.encoder.layers.8.self_attn",
102
  "vision_tower.vision_model.encoder.layers.6.self_attn",
103
- "vision_tower.vision_model.encoder.layers.26.mlp",
 
 
 
104
  "vision_tower.vision_model.encoder.layers.0.self_attn",
 
105
  "vision_tower.vision_model.encoder.layers.0.mlp",
106
- "vision_tower.vision_model.encoder.layers.2.self_attn"
 
107
  ],
108
  "llm_int8_threshold": 6.0,
109
  "load_in_4bit": true,
@@ -141,7 +137,7 @@
141
  "vocab_size": 262208
142
  },
143
  "torch_dtype": "bfloat16",
144
- "transformers_version": "4.50.0.dev0",
145
  "unsloth_fixed": true,
146
  "vision_config": {
147
  "attention_dropout": 0.0,
 
25
  "multi_modal_projector",
26
  "merger",
27
  "modality_projection",
28
+ "language_model.model.layers.12.mlp",
29
+ "language_model.model.layers.3.self_attn",
30
+ "language_model.model.layers.9.mlp",
31
+ "language_model.model.layers.13.self_attn",
32
+ "language_model.model.layers.14.self_attn",
33
+ "language_model.model.layers.4.self_attn",
34
+ "language_model.model.layers.1.mlp",
35
+ "language_model.model.layers.0.mlp",
36
+ "language_model.model.layers.5.self_attn",
37
  "language_model.model.layers.2.mlp",
38
+ "language_model.model.layers.0.self_attn",
39
+ "language_model.model.layers.8.mlp",
40
+ "language_model.model.layers.7.self_attn",
41
+ "language_model.model.layers.6.self_attn",
42
+ "language_model.model.layers.2.self_attn",
43
+ "language_model.model.layers.6.mlp",
44
  "language_model.model.layers.11.mlp",
45
+ "language_model.model.layers.5.mlp",
46
  "language_model.model.layers.4.mlp",
47
+ "vision_tower.vision_model.encoder.layers.25.self_attn",
48
+ "vision_tower.vision_model.encoder.layers.26.self_attn",
 
 
 
 
 
 
 
 
 
 
 
49
  "language_model.model.layers.7.mlp",
50
+ "vision_tower.vision_model.encoder.layers.22.self_attn",
 
 
 
51
  "language_model.model.layers.3.mlp",
 
 
 
 
52
  "vision_tower.vision_model.encoder.layers.23.self_attn",
 
 
 
53
  "vision_tower.vision_model.encoder.layers.19.mlp",
54
+ "vision_tower.vision_model.encoder.layers.15.mlp",
55
+ "vision_tower.vision_model.encoder.layers.25.mlp",
56
  "vision_tower.vision_model.encoder.layers.18.self_attn",
57
+ "vision_tower.vision_model.encoder.layers.21.self_attn",
58
+ "vision_tower.vision_model.encoder.layers.20.self_attn",
59
+ "vision_tower.vision_model.encoder.layers.19.self_attn",
60
+ "vision_tower.vision_model.encoder.layers.24.self_attn",
61
  "vision_tower.vision_model.encoder.layers.16.self_attn",
62
+ "vision_tower.vision_model.encoder.layers.18.mlp",
 
63
  "vision_tower.vision_model.encoder.layers.15.self_attn",
64
+ "vision_tower.vision_model.encoder.layers.17.mlp",
65
+ "vision_tower.vision_model.encoder.layers.17.self_attn",
66
  "vision_tower.vision_model.encoder.layers.13.mlp",
67
+ "vision_tower.vision_model.encoder.layers.20.mlp",
68
+ "vision_tower.vision_model.encoder.layers.12.mlp",
69
+ "vision_tower.vision_model.encoder.layers.22.mlp",
70
  "vision_tower.vision_model.encoder.layers.23.mlp",
71
+ "vision_tower.vision_model.encoder.layers.24.mlp",
 
72
  "vision_tower.vision_model.encoder.layers.16.mlp",
73
+ "vision_tower.vision_model.encoder.layers.10.mlp",
74
+ "vision_tower.vision_model.encoder.layers.11.mlp",
75
  "vision_tower.vision_model.encoder.layers.9.self_attn",
76
+ "vision_tower.vision_model.encoder.layers.14.mlp",
77
+ "vision_tower.vision_model.encoder.layers.14.self_attn",
78
+ "vision_tower.vision_model.encoder.layers.21.mlp",
79
  "vision_tower.vision_model.encoder.layers.12.self_attn",
80
+ "vision_tower.vision_model.encoder.layers.8.self_attn",
 
 
81
  "vision_tower.vision_model.encoder.layers.9.mlp",
 
 
 
 
 
 
82
  "vision_tower.vision_model.encoder.layers.4.self_attn",
83
+ "vision_tower.vision_model.encoder.layers.10.self_attn",
84
+ "vision_tower.vision_model.encoder.layers.7.mlp",
85
  "vision_tower.vision_model.encoder.layers.5.mlp",
86
+ "vision_tower.vision_model.encoder.layers.8.mlp",
87
+ "vision_tower.vision_model.encoder.layers.4.mlp",
88
  "vision_tower.vision_model.encoder.layers.7.self_attn",
89
+ "vision_tower.vision_model.encoder.layers.1.mlp",
90
+ "vision_tower.vision_model.encoder.layers.11.self_attn",
 
91
  "vision_tower.vision_model.encoder.layers.1.self_attn",
92
+ "vision_tower.vision_model.encoder.layers.6.mlp",
93
  "vision_tower.vision_model.encoder.layers.6.self_attn",
94
+ "vision_tower.vision_model.encoder.layers.5.self_attn",
95
+ "vision_tower.vision_model.encoder.layers.13.self_attn",
96
+ "vision_tower.vision_model.encoder.layers.3.mlp",
97
+ "vision_tower.vision_model.encoder.layers.2.mlp",
98
  "vision_tower.vision_model.encoder.layers.0.self_attn",
99
+ "vision_tower.vision_model.encoder.layers.3.self_attn",
100
  "vision_tower.vision_model.encoder.layers.0.mlp",
101
+ "vision_tower.vision_model.encoder.layers.2.self_attn",
102
+ "vision_tower.vision_model.encoder.layers.26.mlp"
103
  ],
104
  "llm_int8_threshold": 6.0,
105
  "load_in_4bit": true,
 
137
  "vocab_size": 262208
138
  },
139
  "torch_dtype": "bfloat16",
140
+ "transformers_version": "4.51.0",
141
  "unsloth_fixed": true,
142
  "vision_config": {
143
  "attention_dropout": 0.0,
generation_config.json CHANGED
@@ -1,8 +1,13 @@
1
  {
2
- "_from_model_config": true,
3
  "bos_token_id": 2,
4
  "cache_implementation": "hybrid",
5
- "eos_token_id": 1,
 
 
 
 
6
  "pad_token_id": 0,
7
- "transformers_version": "4.50.0.dev0"
 
 
8
  }
 
1
  {
 
2
  "bos_token_id": 2,
3
  "cache_implementation": "hybrid",
4
+ "do_sample": true,
5
+ "eos_token_id": [
6
+ 1,
7
+ 106
8
+ ],
9
  "pad_token_id": 0,
10
+ "top_k": 64,
11
+ "top_p": 0.95,
12
+ "transformers_version": "4.51.0"
13
  }
model-00001-of-00003.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:636caa7592c252469d386c7a567c5ef6b34bac754edb95a033721e0ff3021ab9
3
- size 4960125164
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0170076efc281379f6a0c3b789bdb89e11039ead490e7261bb189a1d0704cbaf
3
+ size 4909879823
model-00002-of-00003.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:ed20ac4b90eada6a5c89633f495aad5bd41e0a186421a0ebfed9ac1ccbe64701
3
- size 4992846171
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d97a48759f90c15c1a1d7df00dc4ab2e09f2e6293a7b93cb15d389a42b09ae7b
3
+ size 4991126288
model-00003-of-00003.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:a7bedb8fc3a7b4ea28c590264748a908f868256fc48930c13692606e4e6dab21
3
- size 3007436388
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b1bd94504135f7d3bd95f81af0930f4e6c4ddce4119cc2b430b1f0f15e2c37ef
3
+ size 2289062857
model.safetensors.index.json CHANGED
The diff for this file is too large to render. See raw diff