Hugging Face's logo

Argobell
/
gemma-3n-gek408-dpo

Image-Text-to-Text
Transformers
Safetensors
English
gemma3n
sft
dpo
unsloth
instruction-tuning
text-generation
multimodal
education
reasoning
conversational
Model card Files
xet
Community
Argobell commited on
Commit
805df1d
·
verified ·
1 Parent(s): 3016608

Upload folder using huggingface_hub

Browse files
Files changed (15) hide show
  1. .gitattributes +1 -0
  2. chat_template.jinja +49 -0
  3. config.json +209 -0
  4. configuration.json +1 -0
  5. generation_config.json +13 -0
  6. model-00001-of-00003.safetensors +3 -0
  7. model-00002-of-00003.safetensors +3 -0
  8. model-00003-of-00003.safetensors +3 -0
  9. model.safetensors.index.json +0 -0
  10. preprocessor_config.json +51 -0
  11. processor_config.json +5 -0
  12. special_tokens_map.json +36 -0
  13. tokenizer.json +3 -0
  14. tokenizer.model +3 -0
  15. tokenizer_config.json +0 -0
.gitattributes CHANGED
@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
36
+ tokenizer.json filter=lfs diff=lfs merge=lfs -text
chat_template.jinja ADDED
@@ -0,0 +1,49 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {{ bos_token }}
2
+ {%- if messages[0]['role'] == 'system' -%}
3
+ {%- if messages[0]['content'] is string -%}
4
+ {%- set first_user_prefix = messages[0]['content'] + '
5
+
6
+ ' -%}
7
+ {%- else -%}
8
+ {%- set first_user_prefix = messages[0]['content'][0]['text'] + '
9
+
10
+ ' -%}
11
+ {%- endif -%}
12
+ {%- set loop_messages = messages[1:] -%}
13
+ {%- else -%}
14
+ {%- set first_user_prefix = "" -%}
15
+ {%- set loop_messages = messages -%}
16
+ {%- endif -%}
17
+ {%- for message in loop_messages -%}
18
+ {%- if (message['role'] == 'user') != (loop.index0 % 2 == 0) -%}
19
+ {{ raise_exception("Conversation roles must alternate user/assistant/user/assistant/...") }}
20
+ {%- endif -%}
21
+ {%- if (message['role'] == 'assistant') -%}
22
+ {%- set role = "model" -%}
23
+ {%- else -%}
24
+ {%- set role = message['role'] -%}
25
+ {%- endif -%}
26
+ {{ '<start_of_turn>' + role + '
27
+ ' + (first_user_prefix if loop.first else "") }}
28
+ {%- if message['content'] is string -%}
29
+ {{ message['content'] | trim }}
30
+ {%- elif message['content'] is iterable -%}
31
+ {%- for item in message['content'] -%}
32
+ {%- if item['type'] == 'audio' -%}
33
+ {{ '<audio_soft_token>' }}
34
+ {%- elif item['type'] == 'image' -%}
35
+ {{ '<image_soft_token>' }}
36
+ {%- elif item['type'] == 'text' -%}
37
+ {{ item['text'] | trim }}
38
+ {%- endif -%}
39
+ {%- endfor -%}
40
+ {%- else -%}
41
+ {{ raise_exception("Invalid content type") }}
42
+ {%- endif -%}
43
+ {{ '<end_of_turn>
44
+ ' }}
45
+ {%- endfor -%}
46
+ {%- if add_generation_prompt -%}
47
+ {{'<start_of_turn>model
48
+ '}}
49
+ {%- endif -%}
config.json ADDED
@@ -0,0 +1,209 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Gemma3nForConditionalGeneration"
4
+ ],
5
+ "audio_config": {
6
+ "conf_attention_chunk_size": 12,
7
+ "conf_attention_context_left": 13,
8
+ "conf_attention_context_right": 0,
9
+ "conf_attention_logit_cap": 50.0,
10
+ "conf_conv_kernel_size": 5,
11
+ "conf_num_attention_heads": 8,
12
+ "conf_num_hidden_layers": 12,
13
+ "conf_reduction_factor": 4,
14
+ "conf_residual_weight": 0.5,
15
+ "gradient_clipping": 10000000000.0,
16
+ "hidden_size": 1536,
17
+ "input_feat_size": 128,
18
+ "model_type": "gemma3n_audio",
19
+ "rms_norm_eps": 1e-06,
20
+ "sscp_conv_channel_size": [
21
+ 128,
22
+ 32
23
+ ],
24
+ "sscp_conv_group_norm_eps": 0.001,
25
+ "sscp_conv_kernel_size": [
26
+ [
27
+ 3,
28
+ 3
29
+ ],
30
+ [
31
+ 3,
32
+ 3
33
+ ]
34
+ ],
35
+ "sscp_conv_stride_size": [
36
+ [
37
+ 2,
38
+ 2
39
+ ],
40
+ [
41
+ 2,
42
+ 2
43
+ ]
44
+ ],
45
+ "torch_dtype": "bfloat16",
46
+ "vocab_offset": 262272,
47
+ "vocab_size": 128
48
+ },
49
+ "audio_soft_tokens_per_image": 188,
50
+ "audio_token_id": 262273,
51
+ "boa_token_id": 256000,
52
+ "boi_token_id": 255999,
53
+ "bos_token_id": 2,
54
+ "eoa_token_id": 262272,
55
+ "eoi_token_id": 262144,
56
+ "eos_token_id": 106,
57
+ "image_token_id": 262145,
58
+ "initializer_range": 0.02,
59
+ "model_type": "gemma3n",
60
+ "pad_token_id": 0,
61
+ "text_config": {
62
+ "activation_sparsity_pattern": [
63
+ 0.95,
64
+ 0.95,
65
+ 0.95,
66
+ 0.95,
67
+ 0.95,
68
+ 0.95,
69
+ 0.95,
70
+ 0.95,
71
+ 0.95,
72
+ 0.95,
73
+ 0.0,
74
+ 0.0,
75
+ 0.0,
76
+ 0.0,
77
+ 0.0,
78
+ 0.0,
79
+ 0.0,
80
+ 0.0,
81
+ 0.0,
82
+ 0.0,
83
+ 0.0,
84
+ 0.0,
85
+ 0.0,
86
+ 0.0,
87
+ 0.0,
88
+ 0.0,
89
+ 0.0,
90
+ 0.0,
91
+ 0.0,
92
+ 0.0
93
+ ],
94
+ "altup_active_idx": 0,
95
+ "altup_coef_clip": 120.0,
96
+ "altup_correct_scale": true,
97
+ "altup_num_inputs": 4,
98
+ "attention_bias": false,
99
+ "attention_dropout": 0.0,
100
+ "final_logit_softcapping": 30.0,
101
+ "head_dim": 256,
102
+ "hidden_activation": "gelu_pytorch_tanh",
103
+ "hidden_size": 2048,
104
+ "hidden_size_per_layer_input": 256,
105
+ "initializer_range": 0.02,
106
+ "intermediate_size": [
107
+ 8192,
108
+ 8192,
109
+ 8192,
110
+ 8192,
111
+ 8192,
112
+ 8192,
113
+ 8192,
114
+ 8192,
115
+ 8192,
116
+ 8192,
117
+ 8192,
118
+ 8192,
119
+ 8192,
120
+ 8192,
121
+ 8192,
122
+ 8192,
123
+ 8192,
124
+ 8192,
125
+ 8192,
126
+ 8192,
127
+ 8192,
128
+ 8192,
129
+ 8192,
130
+ 8192,
131
+ 8192,
132
+ 8192,
133
+ 8192,
134
+ 8192,
135
+ 8192,
136
+ 8192
137
+ ],
138
+ "laurel_rank": 64,
139
+ "layer_types": [
140
+ "sliding_attention",
141
+ "sliding_attention",
142
+ "sliding_attention",
143
+ "sliding_attention",
144
+ "full_attention",
145
+ "sliding_attention",
146
+ "sliding_attention",
147
+ "sliding_attention",
148
+ "sliding_attention",
149
+ "full_attention",
150
+ "sliding_attention",
151
+ "sliding_attention",
152
+ "sliding_attention",
153
+ "sliding_attention",
154
+ "full_attention",
155
+ "sliding_attention",
156
+ "sliding_attention",
157
+ "sliding_attention",
158
+ "sliding_attention",
159
+ "full_attention",
160
+ "sliding_attention",
161
+ "sliding_attention",
162
+ "sliding_attention",
163
+ "sliding_attention",
164
+ "full_attention",
165
+ "sliding_attention",
166
+ "sliding_attention",
167
+ "sliding_attention",
168
+ "sliding_attention",
169
+ "full_attention"
170
+ ],
171
+ "max_position_embeddings": 32768,
172
+ "model_type": "gemma3n_text",
173
+ "num_attention_heads": 8,
174
+ "num_hidden_layers": 30,
175
+ "num_key_value_heads": 2,
176
+ "num_kv_shared_layers": 10,
177
+ "rms_norm_eps": 1e-06,
178
+ "rope_local_base_freq": 10000.0,
179
+ "rope_scaling": null,
180
+ "rope_theta": 1000000.0,
181
+ "sliding_window": 512,
182
+ "torch_dtype": "bfloat16",
183
+ "use_cache": true,
184
+ "vocab_size": 262400,
185
+ "vocab_size_per_layer_input": 262144
186
+ },
187
+ "torch_dtype": "bfloat16",
188
+ "transformers_version": "4.54.1",
189
+ "unsloth_fixed": true,
190
+ "unsloth_version": "2025.7.2",
191
+ "vision_config": {
192
+ "architecture": "mobilenetv5_300m_enc",
193
+ "do_pooling": false,
194
+ "hidden_size": 2048,
195
+ "initializer_range": 0.02,
196
+ "label_names": [
197
+ "LABEL_0",
198
+ "LABEL_1"
199
+ ],
200
+ "model_args": null,
201
+ "model_type": "gemma3n_vision",
202
+ "num_classes": 2,
203
+ "rms_norm_eps": 1e-06,
204
+ "torch_dtype": "bfloat16",
205
+ "vocab_offset": 262144,
206
+ "vocab_size": 128
207
+ },
208
+ "vision_soft_tokens_per_image": 256
209
+ }
configuration.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"framework":"Pytorch","task":"image-text-to-text"}
generation_config.json ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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": "4.54.1"
13
+ }
model-00001-of-00003.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1af26e1fd61af0dc067252c907bf52900c7cd5864893e29970e6ea87320322a6
3
+ size 3077103824
model-00002-of-00003.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f93a09621eaba66f4dc05c838969d7e3e8273e61958cbf1c84c79938f540f58c
3
+ size 4981242176
model-00003-of-00003.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bba911c31f0cfcaf9771e79a83aa7646ca31abce06a5e5c3371da49432dda44c
3
+ size 2820739840
model.safetensors.index.json ADDED
The diff for this file is too large to render. See raw diff
 
preprocessor_config.json ADDED
@@ -0,0 +1,51 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "crop_size": null,
3
+ "data_format": "channels_first",
4
+ "default_to_square": false,
5
+ "device": null,
6
+ "disable_grouping": null,
7
+ "dither": 0.0,
8
+ "do_center_crop": null,
9
+ "do_convert_rgb": null,
10
+ "do_normalize": false,
11
+ "do_rescale": true,
12
+ "do_resize": true,
13
+ "feature_size": 128,
14
+ "fft_length": 1024,
15
+ "fft_overdrive": true,
16
+ "frame_length": 512,
17
+ "hop_length": 160,
18
+ "image_mean": [
19
+ 0.5,
20
+ 0.5,
21
+ 0.5
22
+ ],
23
+ "image_processor_type": "SiglipImageProcessorFast",
24
+ "image_seq_length": 256,
25
+ "image_std": [
26
+ 0.5,
27
+ 0.5,
28
+ 0.5
29
+ ],
30
+ "input_data_format": null,
31
+ "input_scale_factor": 1.0,
32
+ "max_frequency": 7600.0,
33
+ "mel_floor": 1e-05,
34
+ "min_frequency": 125.0,
35
+ "padding_side": "right",
36
+ "padding_value": 0.0,
37
+ "per_bin_mean": null,
38
+ "per_bin_stddev": null,
39
+ "preemphasis": 0.97,
40
+ "preemphasis_htk_flavor": true,
41
+ "processor_class": "Gemma3nProcessor",
42
+ "resample": 2,
43
+ "rescale_factor": 0.00392156862745098,
44
+ "return_attention_mask": true,
45
+ "return_tensors": null,
46
+ "sampling_rate": 16000,
47
+ "size": {
48
+ "height": 768,
49
+ "width": 768
50
+ }
51
+ }
processor_config.json ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ {
2
+ "audio_seq_length": 188,
3
+ "image_seq_length": 256,
4
+ "processor_class": "Gemma3nProcessor"
5
+ }
special_tokens_map.json ADDED
@@ -0,0 +1,36 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "audio_token": "<audio_soft_token>",
3
+ "boa_token": "<start_of_audio>",
4
+ "boi_token": "<start_of_image>",
5
+ "bos_token": {
6
+ "content": "<bos>",
7
+ "lstrip": false,
8
+ "normalized": false,
9
+ "rstrip": false,
10
+ "single_word": false
11
+ },
12
+ "eoa_token": "<end_of_audio>",
13
+ "eoi_token": "<end_of_image>",
14
+ "eos_token": {
15
+ "content": "<end_of_turn>",
16
+ "lstrip": false,
17
+ "normalized": false,
18
+ "rstrip": false,
19
+ "single_word": false
20
+ },
21
+ "image_token": "<image_soft_token>",
22
+ "pad_token": {
23
+ "content": "<pad>",
24
+ "lstrip": false,
25
+ "normalized": false,
26
+ "rstrip": false,
27
+ "single_word": false
28
+ },
29
+ "unk_token": {
30
+ "content": "<unk>",
31
+ "lstrip": false,
32
+ "normalized": false,
33
+ "rstrip": false,
34
+ "single_word": false
35
+ }
36
+ }
tokenizer.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:34e3f436416642ff2b4cba0602975a87bd8766c9f3b1851dde39479073bafd08
3
+ size 33442807
tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ea5f0cc48abfbfc04d14562270a32e02149a3e7035f368cc5a462786f4a59961
3
+ size 4696020
tokenizer_config.json ADDED
The diff for this file is too large to render. See raw diff