Hugging Face's logo

lmstudio-community
/
gemma-3n-E4B-it-MLX-bf16

Image-Text-to-Text
Transformers
Safetensors
MLX
gemma3n
automatic-speech-recognition
automatic-speech-translation
audio-text-to-text
video-text-to-text
conversational
Model card Files
xet
Community
1
lmmy commited on
Commit
b531c7c
·
verified ·
1 Parent(s): 327421f

Add files using upload-large-folder tool

Browse files
Files changed (11) hide show
  1. README.md +17 -7
  2. chat_template.jinja +49 -0
  3. config.json +340 -81
  4. generation_config.json +1 -1
  5. model-00001-of-00004.safetensors +2 -2
  6. model-00002-of-00004.safetensors +2 -2
  7. model-00003-of-00004.safetensors +2 -2
  8. model-00004-of-00004.safetensors +2 -2
  9. model.safetensors.index.json +2 -2
  10. preprocessor_config.json +3 -3
  11. tokenizer_config.json +33 -33
README.md CHANGED
@@ -1,26 +1,36 @@
1
  ---
 
 
 
 
 
 
 
 
2
  base_model: google/gemma-3n-E4B-it
3
  tags:
 
 
 
 
4
  - mlx
5
  ---
6
- ## 💫 Community Model> gemma-3n-E4B-it by Google
7
 
8
- _👾 [LM Studio](https://lmstudio.ai) Community models highlights program. Highlighting new & noteworthy models by the community. Join the conversation on [Discord](https://discord.gg/aPQfnNkxGC)_.
9
 
10
- **Model creator:** [Google](https://huggingface.co/Google)<br>
11
  **Original model**: [gemma-3n-E4B-it](https://huggingface.co/google/gemma-3n-E4B-it)<br>
12
- **MLX conversion:** provided by [LM Studio team](https://x.com/lmstudio) using [mlx_vlm](https://github.com/Blaizzy/mlx-vlm)<br>
13
 
14
  ## Technical Details
15
 
16
  Original bfloat16 version of gemma-3n-E4B-it using MLX, optimized for Apple Silicon.
17
 
18
- Quantizations coming soon.
19
-
20
  ## Special thanks
21
 
22
  🙏 Special thanks to the [Apple Machine Learning Research](https://github.com/ml-explore) team for creating [MLX](https://github.com/ml-explore/mlx).
23
 
24
  ## Disclaimers
25
 
26
- LM Studio is not the creator, originator, or owner of any Model featured in the Community Model Program. Each Community Model is created and provided by third parties. LM Studio does not endorse, support, represent or guarantee the completeness, truthfulness, accuracy, or reliability of any Community Model. You understand that Community Models can produce content that might be offensive, harmful, inaccurate or otherwise inappropriate, or deceptive. Each Community Model is the sole responsibility of the person or entity who originated such Model. LM Studio may not monitor or control the Community Models and cannot, and does not, take responsibility for any such Model. LM Studio disclaims all warranties or guarantees about the accuracy, reliability or benefits of the Community Models. LM Studio further disclaims any warranty that the Community Model will meet your requirements, be secure, uninterrupted or available at any time or location, or error-free, viruses-free, or that any errors will be corrected, or otherwise. You will be solely responsible for any damage resulting from your use of or access to the Community Models, your downloading of any Community Model, or use of any other Community Model provided by or through LM Studio.
 
1
  ---
2
+ license: gemma
3
+ library_name: transformers
4
+ pipeline_tag: image-text-to-text
5
+ extra_gated_heading: Access Gemma on Hugging Face
6
+ extra_gated_prompt: To access Gemma on Hugging Face, you’re required to review and
7
+ agree to Google’s usage license. To do this, please ensure you’re logged in to Hugging
8
+ Face and click below. Requests are processed immediately.
9
+ extra_gated_button_content: Acknowledge license
10
  base_model: google/gemma-3n-E4B-it
11
  tags:
12
+ - automatic-speech-recognition
13
+ - automatic-speech-translation
14
+ - audio-text-to-text
15
+ - video-text-to-text
16
  - mlx
17
  ---
18
+ ## 💫 Community Model> gemma-3n-E4B-it by google
19
 
20
+ *👾 [LM Studio](https://lmstudio.ai) Community models highlights program. Highlighting new & noteworthy models by the community. Join the conversation on [Discord](https://discord.gg/aPQfnNkxGC)*.
21
 
22
+ **Model creator:** [google](https://huggingface.co/google)<br>
23
  **Original model**: [gemma-3n-E4B-it](https://huggingface.co/google/gemma-3n-E4B-it)<br>
24
+ **MLX quantization:** provided by [LM Studio team](https://x.com/lmstudio) using [mlx_vlm](https://github.com/Blaizzy/mlx-vlm)<br>
25
 
26
  ## Technical Details
27
 
28
  Original bfloat16 version of gemma-3n-E4B-it using MLX, optimized for Apple Silicon.
29
 
 
 
30
  ## Special thanks
31
 
32
  🙏 Special thanks to the [Apple Machine Learning Research](https://github.com/ml-explore) team for creating [MLX](https://github.com/ml-explore/mlx).
33
 
34
  ## Disclaimers
35
 
36
+ LM Studio is not the creator, originator, or owner of any Model featured in the Community Model Program. Each Community Model is created and provided by third parties. LM Studio does not endorse, support, represent or guarantee the completeness, truthfulness, accuracy, or reliability of any Community Model. You understand that Community Models can produce content that might be offensive, harmful, inaccurate or otherwise inappropriate, or deceptive. Each Community Model is the sole responsibility of the person or entity who originated such Model. LM Studio may not monitor or control the Community Models and cannot, and does not, take responsibility for any such Model. LM Studio disclaims all warranties or guarantees about the accuracy, reliability or benefits of the Community Models. LM Studio further disclaims any warranty that the Community Model will meet your requirements, be secure, uninterrupted or available at any time or location, or error-free, viruses-free, or that any errors will be corrected, or otherwise. You will be solely responsible for any damage resulting from your use of or access to the Community Models, your downloading of any Community Model, or use of any other Community Model provided by or through LM Studio.
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 CHANGED
@@ -1,28 +1,89 @@
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_positional_bias_size": 256,
14
  "conf_reduction_factor": 4,
15
  "conf_residual_weight": 0.5,
16
- "gradient_clipping": 10000000000.0,
17
- "hidden_size": 1536,
18
- "input_feat_size": 128,
19
- "model_type": "gemma3n_audio",
20
- "rms_norm_eps": 1e-06,
21
  "sscp_conv_channel_size": [
22
  128,
23
  32
24
  ],
25
- "sscp_conv_eps": 0.001,
26
  "sscp_conv_kernel_size": [
27
  [
28
  3,
@@ -43,76 +104,182 @@
43
  2
44
  ]
45
  ],
46
- "torch_dtype": "bfloat16",
47
- "vocab_offset": 262272,
48
- "vocab_size": 128
49
  },
50
  "audio_soft_tokens_per_image": 188,
51
  "audio_token_id": 262273,
 
 
52
  "boa_token_id": 256000,
53
  "boi_token_id": 255999,
 
 
 
 
 
 
 
 
54
  "eoa_token_id": 262272,
55
  "eoi_token_id": 262144,
56
  "eos_token_id": [
57
  1,
58
  106
59
  ],
 
 
 
 
 
 
 
 
60
  "image_token_id": 262145,
61
  "initializer_range": 0.02,
 
 
 
 
 
 
 
 
 
62
  "model_type": "gemma3n",
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
63
  "text_config": {
64
- "activation_sparsity_pattern": [
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
- 0.95,
74
- 0.95,
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
- 0.0,
94
- 0.0,
95
- 0.0,
96
- 0.0,
97
- 0.0,
98
- 0.0,
99
- 0.0
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
100
  ],
101
- "altup_active_idx": 0,
102
- "altup_coef_clip": 120.0,
103
- "altup_correct_scale": true,
104
- "altup_lr_multiplier": 1.0,
105
- "altup_num_inputs": 4,
 
 
 
106
  "attention_bias": false,
107
  "attention_dropout": 0.0,
108
- "final_logit_softcapping": 30.0,
109
- "head_dim": 256,
110
  "hidden_activation": "gelu_pytorch_tanh",
111
- "hidden_size": 2048,
112
- "hidden_size_per_layer_input": 256,
113
- "initializer_range": 0.02,
114
- "intermediate_size": 16384,
115
- "laurel_rank": 64,
116
  "layer_types": [
117
  "sliding_attention",
118
  "sliding_attention",
@@ -150,39 +317,131 @@
150
  "sliding_attention",
151
  "full_attention"
152
  ],
153
- "max_position_embeddings": 32768,
154
- "model_type": "gemma3n_text",
155
- "num_attention_heads": 8,
156
- "num_hidden_layers": 35,
157
- "num_key_value_heads": 2,
158
- "num_kv_shared_layers": 15,
159
- "query_pre_attn_scalar": 256,
160
- "rms_norm_eps": 1e-06,
161
  "rope_local_base_freq": 10000.0,
162
  "rope_scaling": null,
163
- "rope_theta": 1000000.0,
164
- "sliding_window": 512,
165
- "torch_dtype": "bfloat16",
166
- "use_cache": true,
167
- "vocab_size": 262400,
168
- "vocab_size_per_layer_input": 262144
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
169
  },
170
- "transformers_version": "4.53.0.dev0",
 
 
 
 
 
 
 
 
 
171
  "vision_config": {
172
- "architecture": "mobilenetv5_300m_enc",
173
- "do_pooling": true,
174
- "hidden_size": 2048,
175
- "initializer_range": 0.02,
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
176
  "label_names": [
177
  "LABEL_0",
178
  "LABEL_1"
179
  ],
180
  "model_type": "gemma3n_vision",
181
  "num_classes": 2,
182
- "rms_norm_eps": 1e-06,
183
- "torch_dtype": "bfloat16",
 
 
 
 
184
  "vocab_offset": 262144,
185
- "vocab_size": 128
 
186
  },
187
  "vision_soft_tokens_per_image": 256
188
  }
 
1
  {
2
+ "add_cross_attention": false,
3
  "architectures": [
4
  "Gemma3nForConditionalGeneration"
5
  ],
6
  "audio_config": {
7
+ "return_dict": true,
8
+ "output_hidden_states": false,
9
+ "torchscript": false,
10
+ "torch_dtype": "bfloat16",
11
+ "use_bfloat16": false,
12
+ "tf_legacy_loss": false,
13
+ "pruned_heads": {},
14
+ "tie_word_embeddings": true,
15
+ "chunk_size_feed_forward": 0,
16
+ "is_encoder_decoder": false,
17
+ "is_decoder": false,
18
+ "cross_attention_hidden_size": null,
19
+ "add_cross_attention": false,
20
+ "tie_encoder_decoder": false,
21
+ "max_length": 20,
22
+ "min_length": 0,
23
+ "do_sample": false,
24
+ "early_stopping": false,
25
+ "num_beams": 1,
26
+ "num_beam_groups": 1,
27
+ "diversity_penalty": 0.0,
28
+ "temperature": 1.0,
29
+ "top_k": 50,
30
+ "top_p": 1.0,
31
+ "typical_p": 1.0,
32
+ "repetition_penalty": 1.0,
33
+ "length_penalty": 1.0,
34
+ "no_repeat_ngram_size": 0,
35
+ "encoder_no_repeat_ngram_size": 0,
36
+ "bad_words_ids": null,
37
+ "num_return_sequences": 1,
38
+ "output_scores": false,
39
+ "return_dict_in_generate": false,
40
+ "forced_bos_token_id": null,
41
+ "forced_eos_token_id": null,
42
+ "remove_invalid_values": false,
43
+ "exponential_decay_length_penalty": null,
44
+ "suppress_tokens": null,
45
+ "begin_suppress_tokens": null,
46
+ "architectures": null,
47
+ "finetuning_task": null,
48
+ "id2label": {
49
+ "0": "LABEL_0",
50
+ "1": "LABEL_1"
51
+ },
52
+ "label2id": {
53
+ "LABEL_0": 0,
54
+ "LABEL_1": 1
55
+ },
56
+ "tokenizer_class": null,
57
+ "prefix": null,
58
+ "bos_token_id": null,
59
+ "pad_token_id": null,
60
+ "eos_token_id": null,
61
+ "sep_token_id": null,
62
+ "decoder_start_token_id": null,
63
+ "task_specific_params": null,
64
+ "problem_type": null,
65
+ "_name_or_path": "",
66
+ "model_type": "gemma3n_audio",
67
+ "input_feat_size": 128,
68
+ "hidden_size": 1536,
69
+ "rms_norm_eps": 1e-06,
70
+ "vocab_size": 128,
71
+ "vocab_offset": 262272,
72
+ "gradient_clipping": 10000000000.0,
73
  "conf_attention_chunk_size": 12,
74
  "conf_attention_context_left": 13,
75
  "conf_attention_context_right": 0,
76
  "conf_attention_logit_cap": 50.0,
 
77
  "conf_num_attention_heads": 8,
78
  "conf_num_hidden_layers": 12,
79
+ "conf_conv_kernel_size": 5,
80
  "conf_reduction_factor": 4,
81
  "conf_residual_weight": 0.5,
 
 
 
 
 
82
  "sscp_conv_channel_size": [
83
  128,
84
  32
85
  ],
86
+ "sscp_conv_group_norm_eps": 0.001,
87
  "sscp_conv_kernel_size": [
88
  [
89
  3,
 
104
  2
105
  ]
106
  ],
107
+ "output_attentions": false
 
 
108
  },
109
  "audio_soft_tokens_per_image": 188,
110
  "audio_token_id": 262273,
111
+ "bad_words_ids": null,
112
+ "begin_suppress_tokens": null,
113
  "boa_token_id": 256000,
114
  "boi_token_id": 255999,
115
+ "bos_token_id": null,
116
+ "chunk_size_feed_forward": 0,
117
+ "cross_attention_hidden_size": null,
118
+ "decoder_start_token_id": null,
119
+ "diversity_penalty": 0.0,
120
+ "do_sample": false,
121
+ "early_stopping": false,
122
+ "encoder_no_repeat_ngram_size": 0,
123
  "eoa_token_id": 262272,
124
  "eoi_token_id": 262144,
125
  "eos_token_id": [
126
  1,
127
  106
128
  ],
129
+ "exponential_decay_length_penalty": null,
130
+ "finetuning_task": null,
131
+ "forced_bos_token_id": null,
132
+ "forced_eos_token_id": null,
133
+ "id2label": {
134
+ "0": "LABEL_0",
135
+ "1": "LABEL_1"
136
+ },
137
  "image_token_id": 262145,
138
  "initializer_range": 0.02,
139
+ "is_decoder": false,
140
+ "is_encoder_decoder": false,
141
+ "label2id": {
142
+ "LABEL_0": 0,
143
+ "LABEL_1": 1
144
+ },
145
+ "length_penalty": 1.0,
146
+ "max_length": 20,
147
+ "min_length": 0,
148
  "model_type": "gemma3n",
149
+ "no_repeat_ngram_size": 0,
150
+ "num_beam_groups": 1,
151
+ "num_beams": 1,
152
+ "num_return_sequences": 1,
153
+ "output_attentions": false,
154
+ "output_hidden_states": false,
155
+ "output_scores": false,
156
+ "pad_token_id": null,
157
+ "prefix": null,
158
+ "problem_type": null,
159
+ "pruned_heads": {},
160
+ "remove_invalid_values": false,
161
+ "repetition_penalty": 1.0,
162
+ "return_dict": true,
163
+ "return_dict_in_generate": false,
164
+ "sep_token_id": null,
165
+ "suppress_tokens": null,
166
+ "task_specific_params": null,
167
+ "temperature": 1.0,
168
  "text_config": {
169
+ "return_dict": true,
170
+ "output_hidden_states": false,
171
+ "torchscript": false,
172
+ "torch_dtype": "bfloat16",
173
+ "use_bfloat16": false,
174
+ "tf_legacy_loss": false,
175
+ "pruned_heads": {},
176
+ "tie_word_embeddings": true,
177
+ "chunk_size_feed_forward": 0,
178
+ "is_encoder_decoder": false,
179
+ "is_decoder": false,
180
+ "cross_attention_hidden_size": null,
181
+ "add_cross_attention": false,
182
+ "tie_encoder_decoder": false,
183
+ "max_length": 20,
184
+ "min_length": 0,
185
+ "do_sample": false,
186
+ "early_stopping": false,
187
+ "num_beams": 1,
188
+ "num_beam_groups": 1,
189
+ "diversity_penalty": 0.0,
190
+ "temperature": 1.0,
191
+ "top_k": 50,
192
+ "top_p": 1.0,
193
+ "typical_p": 1.0,
194
+ "repetition_penalty": 1.0,
195
+ "length_penalty": 1.0,
196
+ "no_repeat_ngram_size": 0,
197
+ "encoder_no_repeat_ngram_size": 0,
198
+ "bad_words_ids": null,
199
+ "num_return_sequences": 1,
200
+ "output_scores": false,
201
+ "return_dict_in_generate": false,
202
+ "forced_bos_token_id": null,
203
+ "forced_eos_token_id": null,
204
+ "remove_invalid_values": false,
205
+ "exponential_decay_length_penalty": null,
206
+ "suppress_tokens": null,
207
+ "begin_suppress_tokens": null,
208
+ "architectures": null,
209
+ "finetuning_task": null,
210
+ "id2label": {
211
+ "0": "LABEL_0",
212
+ "1": "LABEL_1"
213
+ },
214
+ "label2id": {
215
+ "LABEL_0": 0,
216
+ "LABEL_1": 1
217
+ },
218
+ "tokenizer_class": null,
219
+ "prefix": null,
220
+ "bos_token_id": 2,
221
+ "pad_token_id": 0,
222
+ "eos_token_id": 1,
223
+ "sep_token_id": null,
224
+ "decoder_start_token_id": null,
225
+ "task_specific_params": null,
226
+ "problem_type": null,
227
+ "_name_or_path": "",
228
+ "model_type": "gemma3n_text",
229
+ "vocab_size": 262400,
230
+ "vocab_size_per_layer_input": 262144,
231
+ "max_position_embeddings": 32768,
232
+ "hidden_size": 2048,
233
+ "intermediate_size": [
234
+ 16384,
235
+ 16384,
236
+ 16384,
237
+ 16384,
238
+ 16384,
239
+ 16384,
240
+ 16384,
241
+ 16384,
242
+ 16384,
243
+ 16384,
244
+ 16384,
245
+ 16384,
246
+ 16384,
247
+ 16384,
248
+ 16384,
249
+ 16384,
250
+ 16384,
251
+ 16384,
252
+ 16384,
253
+ 16384,
254
+ 16384,
255
+ 16384,
256
+ 16384,
257
+ 16384,
258
+ 16384,
259
+ 16384,
260
+ 16384,
261
+ 16384,
262
+ 16384,
263
+ 16384,
264
+ 16384,
265
+ 16384,
266
+ 16384,
267
+ 16384,
268
+ 16384
269
  ],
270
+ "num_hidden_layers": 35,
271
+ "num_attention_heads": 8,
272
+ "head_dim": 256,
273
+ "num_key_value_heads": 2,
274
+ "initializer_range": 0.02,
275
+ "rms_norm_eps": 1e-06,
276
+ "use_cache": true,
277
+ "rope_theta": 1000000.0,
278
  "attention_bias": false,
279
  "attention_dropout": 0.0,
 
 
280
  "hidden_activation": "gelu_pytorch_tanh",
281
+ "sliding_window": 512,
282
+ "final_logit_softcapping": 30.0,
 
 
 
283
  "layer_types": [
284
  "sliding_attention",
285
  "sliding_attention",
 
317
  "sliding_attention",
318
  "full_attention"
319
  ],
 
 
 
 
 
 
 
 
320
  "rope_local_base_freq": 10000.0,
321
  "rope_scaling": null,
322
+ "hidden_size_per_layer_input": 256,
323
+ "num_kv_shared_layers": 15,
324
+ "altup_active_idx": 0,
325
+ "altup_coef_clip": 120.0,
326
+ "altup_correct_scale": true,
327
+ "altup_num_inputs": 4,
328
+ "laurel_rank": 64,
329
+ "activation_sparsity_pattern": [
330
+ 0.95,
331
+ 0.95,
332
+ 0.95,
333
+ 0.95,
334
+ 0.95,
335
+ 0.95,
336
+ 0.95,
337
+ 0.95,
338
+ 0.95,
339
+ 0.95,
340
+ 0.0,
341
+ 0.0,
342
+ 0.0,
343
+ 0.0,
344
+ 0.0,
345
+ 0.0,
346
+ 0.0,
347
+ 0.0,
348
+ 0.0,
349
+ 0.0,
350
+ 0.0,
351
+ 0.0,
352
+ 0.0,
353
+ 0.0,
354
+ 0.0,
355
+ 0.0,
356
+ 0.0,
357
+ 0.0,
358
+ 0.0,
359
+ 0.0,
360
+ 0.0,
361
+ 0.0,
362
+ 0.0,
363
+ 0.0,
364
+ 0.0
365
+ ],
366
+ "output_attentions": false
367
  },
368
+ "tf_legacy_loss": false,
369
+ "tie_encoder_decoder": false,
370
+ "tie_word_embeddings": true,
371
+ "tokenizer_class": null,
372
+ "top_k": 50,
373
+ "top_p": 1.0,
374
+ "torchscript": false,
375
+ "transformers_version": "4.53.2",
376
+ "typical_p": 1.0,
377
+ "use_bfloat16": false,
378
  "vision_config": {
379
+ "return_dict": true,
380
+ "output_hidden_states": false,
381
+ "torchscript": false,
382
+ "torch_dtype": "bfloat16",
383
+ "use_bfloat16": false,
384
+ "tf_legacy_loss": false,
385
+ "pruned_heads": {},
386
+ "tie_word_embeddings": true,
387
+ "chunk_size_feed_forward": 0,
388
+ "is_encoder_decoder": false,
389
+ "is_decoder": false,
390
+ "cross_attention_hidden_size": null,
391
+ "add_cross_attention": false,
392
+ "tie_encoder_decoder": false,
393
+ "max_length": 20,
394
+ "min_length": 0,
395
+ "do_sample": false,
396
+ "early_stopping": false,
397
+ "num_beams": 1,
398
+ "num_beam_groups": 1,
399
+ "diversity_penalty": 0.0,
400
+ "temperature": 1.0,
401
+ "top_k": 50,
402
+ "top_p": 1.0,
403
+ "typical_p": 1.0,
404
+ "repetition_penalty": 1.0,
405
+ "length_penalty": 1.0,
406
+ "no_repeat_ngram_size": 0,
407
+ "encoder_no_repeat_ngram_size": 0,
408
+ "bad_words_ids": null,
409
+ "num_return_sequences": 1,
410
+ "output_scores": false,
411
+ "return_dict_in_generate": false,
412
+ "forced_bos_token_id": null,
413
+ "forced_eos_token_id": null,
414
+ "remove_invalid_values": false,
415
+ "exponential_decay_length_penalty": null,
416
+ "suppress_tokens": null,
417
+ "begin_suppress_tokens": null,
418
+ "architectures": null,
419
+ "finetuning_task": null,
420
+ "tokenizer_class": null,
421
+ "prefix": null,
422
+ "bos_token_id": null,
423
+ "pad_token_id": null,
424
+ "eos_token_id": null,
425
+ "sep_token_id": null,
426
+ "decoder_start_token_id": null,
427
+ "task_specific_params": null,
428
+ "problem_type": null,
429
+ "_name_or_path": "",
430
  "label_names": [
431
  "LABEL_0",
432
  "LABEL_1"
433
  ],
434
  "model_type": "gemma3n_vision",
435
  "num_classes": 2,
436
+ "initializer_range": 0.02,
437
+ "do_pooling": false,
438
+ "model_args": null,
439
+ "architecture": "mobilenetv5_300m_enc",
440
+ "hidden_size": 2048,
441
+ "vocab_size": 128,
442
  "vocab_offset": 262144,
443
+ "rms_norm_eps": 1e-06,
444
+ "output_attentions": false
445
  },
446
  "vision_soft_tokens_per_image": 256
447
  }
generation_config.json CHANGED
@@ -9,5 +9,5 @@
9
  "pad_token_id": 0,
10
  "top_k": 64,
11
  "top_p": 0.95,
12
- "transformers_version": "4.53.0.dev0"
13
  }
 
9
  "pad_token_id": 0,
10
  "top_k": 64,
11
  "top_p": 0.95,
12
+ "transformers_version": "4.54.0.dev0"
13
  }
model-00001-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:aa377d4a57e6372686a841a3f186f8422cd441bcf0c8a0142ad0f579c549234d
3
- size 5361861720
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:235f81e5bf893ad0883e8d49a8799c16a91aaa5844677537c9a17f6cd77afef8
3
+ size 5361861852
model-00002-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:dad500c2838dc7031ab8ad02cc0f0534db4b1203fbe2c11a9180208b65c0dae5
3
- size 3586985605
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a350248e3de8895cbc25e0ba04859dbd30994710d050d835c7368d110bcd35cd
3
+ size 3586985631
model-00003-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:6cb6afa6dd36eed84679e33f9c71dff0d42cdf9e10391b4fe20076b9a34b15bb
3
- size 4784658158
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ec17170025dccf5b8454cdf495824c8394909fa8e3be50c66356316582c1a032
3
+ size 5357246856
model-00004-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:4a5c3c9748656a72faa7faa1d3b8a5c572ff5699be8881b986a2aaa9443e8974
3
- size 3041461107
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:abe3afa46a2716431208c7e135788596166dcc12654d8cca929b2112ab3dc5a0
3
+ size 1394082206
model.safetensors.index.json CHANGED
@@ -1,7 +1,6 @@
1
  {
2
  "metadata": {
3
- "total_parameters": 8387373328,
4
- "total_size": 15699956256
5
  },
6
  "weight_map": {
7
  "model.audio_tower.conformer.0.attention.attn.k_proj.weight": "model-00001-of-00004.safetensors",
@@ -1673,6 +1672,7 @@
1673
  "model.vision_tower.timm_model.blocks.3.9.layer_scale.gamma": "model-00001-of-00004.safetensors",
1674
  "model.vision_tower.timm_model.blocks.3.9.norm.weight": "model-00001-of-00004.safetensors",
1675
  "model.vision_tower.timm_model.conv_stem.bn.weight": "model-00001-of-00004.safetensors",
 
1676
  "model.vision_tower.timm_model.conv_stem.conv.weight": "model-00001-of-00004.safetensors",
1677
  "model.vision_tower.timm_model.msfa.ffn.pw_exp.bn.weight": "model-00001-of-00004.safetensors",
1678
  "model.vision_tower.timm_model.msfa.ffn.pw_exp.conv.weight": "model-00001-of-00004.safetensors",
 
1
  {
2
  "metadata": {
3
+ "total_size": 15699956384
 
4
  },
5
  "weight_map": {
6
  "model.audio_tower.conformer.0.attention.attn.k_proj.weight": "model-00001-of-00004.safetensors",
 
1672
  "model.vision_tower.timm_model.blocks.3.9.layer_scale.gamma": "model-00001-of-00004.safetensors",
1673
  "model.vision_tower.timm_model.blocks.3.9.norm.weight": "model-00001-of-00004.safetensors",
1674
  "model.vision_tower.timm_model.conv_stem.bn.weight": "model-00001-of-00004.safetensors",
1675
+ "model.vision_tower.timm_model.conv_stem.conv.bias": "model-00001-of-00004.safetensors",
1676
  "model.vision_tower.timm_model.conv_stem.conv.weight": "model-00001-of-00004.safetensors",
1677
  "model.vision_tower.timm_model.msfa.ffn.pw_exp.bn.weight": "model-00001-of-00004.safetensors",
1678
  "model.vision_tower.timm_model.msfa.ffn.pw_exp.conv.weight": "model-00001-of-00004.safetensors",
preprocessor_config.json CHANGED
@@ -3,13 +3,13 @@
3
  "data_format": "channels_first",
4
  "default_to_square": false,
5
  "device": null,
 
6
  "dither": 0.0,
7
  "do_center_crop": null,
8
  "do_convert_rgb": null,
9
  "do_normalize": false,
10
  "do_rescale": true,
11
  "do_resize": true,
12
- "feature_extractor_type": "Gemma3nAudioFeatureExtractor",
13
  "feature_size": 128,
14
  "fft_length": 1024,
15
  "fft_overdrive": true,
@@ -20,7 +20,7 @@
20
  0.5,
21
  0.5
22
  ],
23
- "image_processor_type": "SiglipImageProcessorFast",
24
  "image_seq_length": 256,
25
  "image_std": [
26
  0.5,
@@ -41,7 +41,7 @@
41
  "processor_class": "Gemma3nProcessor",
42
  "resample": 2,
43
  "rescale_factor": 0.00392156862745098,
44
- "return_attention_mask": false,
45
  "return_tensors": null,
46
  "sampling_rate": 16000,
47
  "size": {
 
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,
 
20
  0.5,
21
  0.5
22
  ],
23
+ "image_processor_type": "SiglipImageProcessor",
24
  "image_seq_length": 256,
25
  "image_std": [
26
  0.5,
 
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": {
tokenizer_config.json CHANGED
@@ -1107,7 +1107,7 @@
1107
  "special": false
1108
  },
1109
  "138": {
1110
- "content": "▁▁",
1111
  "lstrip": false,
1112
  "normalized": false,
1113
  "rstrip": false,
@@ -1115,7 +1115,7 @@
1115
  "special": false
1116
  },
1117
  "139": {
1118
- "content": "▁▁▁",
1119
  "lstrip": false,
1120
  "normalized": false,
1121
  "rstrip": false,
@@ -1123,7 +1123,7 @@
1123
  "special": false
1124
  },
1125
  "140": {
1126
- "content": "▁▁▁▁",
1127
  "lstrip": false,
1128
  "normalized": false,
1129
  "rstrip": false,
@@ -1131,7 +1131,7 @@
1131
  "special": false
1132
  },
1133
  "141": {
1134
- "content": "▁▁▁▁▁",
1135
  "lstrip": false,
1136
  "normalized": false,
1137
  "rstrip": false,
@@ -1139,7 +1139,7 @@
1139
  "special": false
1140
  },
1141
  "142": {
1142
- "content": "▁▁▁▁▁▁",
1143
  "lstrip": false,
1144
  "normalized": false,
1145
  "rstrip": false,
@@ -1147,7 +1147,7 @@
1147
  "special": false
1148
  },
1149
  "143": {
1150
- "content": "▁▁▁▁▁▁▁",
1151
  "lstrip": false,
1152
  "normalized": false,
1153
  "rstrip": false,
@@ -1155,7 +1155,7 @@
1155
  "special": false
1156
  },
1157
  "144": {
1158
- "content": "▁▁▁▁▁▁▁▁",
1159
  "lstrip": false,
1160
  "normalized": false,
1161
  "rstrip": false,
@@ -1163,7 +1163,7 @@
1163
  "special": false
1164
  },
1165
  "145": {
1166
- "content": "▁▁▁▁▁▁▁▁▁",
1167
  "lstrip": false,
1168
  "normalized": false,
1169
  "rstrip": false,
@@ -1171,7 +1171,7 @@
1171
  "special": false
1172
  },
1173
  "146": {
1174
- "content": "▁▁▁▁▁▁▁▁▁▁",
1175
  "lstrip": false,
1176
  "normalized": false,
1177
  "rstrip": false,
@@ -1179,7 +1179,7 @@
1179
  "special": false
1180
  },
1181
  "147": {
1182
- "content": "▁▁▁▁▁▁▁▁▁▁▁",
1183
  "lstrip": false,
1184
  "normalized": false,
1185
  "rstrip": false,
@@ -1187,7 +1187,7 @@
1187
  "special": false
1188
  },
1189
  "148": {
1190
- "content": "▁▁▁▁▁▁▁▁▁▁▁▁",
1191
  "lstrip": false,
1192
  "normalized": false,
1193
  "rstrip": false,
@@ -1195,7 +1195,7 @@
1195
  "special": false
1196
  },
1197
  "149": {
1198
- "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁",
1199
  "lstrip": false,
1200
  "normalized": false,
1201
  "rstrip": false,
@@ -1203,7 +1203,7 @@
1203
  "special": false
1204
  },
1205
  "150": {
1206
- "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
1207
  "lstrip": false,
1208
  "normalized": false,
1209
  "rstrip": false,
@@ -1211,7 +1211,7 @@
1211
  "special": false
1212
  },
1213
  "151": {
1214
- "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
1215
  "lstrip": false,
1216
  "normalized": false,
1217
  "rstrip": false,
@@ -1219,7 +1219,7 @@
1219
  "special": false
1220
  },
1221
  "152": {
1222
- "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
1223
  "lstrip": false,
1224
  "normalized": false,
1225
  "rstrip": false,
@@ -1227,7 +1227,7 @@
1227
  "special": false
1228
  },
1229
  "153": {
1230
- "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
1231
  "lstrip": false,
1232
  "normalized": false,
1233
  "rstrip": false,
@@ -1235,7 +1235,7 @@
1235
  "special": false
1236
  },
1237
  "154": {
1238
- "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
1239
  "lstrip": false,
1240
  "normalized": false,
1241
  "rstrip": false,
@@ -1243,7 +1243,7 @@
1243
  "special": false
1244
  },
1245
  "155": {
1246
- "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
1247
  "lstrip": false,
1248
  "normalized": false,
1249
  "rstrip": false,
@@ -1251,7 +1251,7 @@
1251
  "special": false
1252
  },
1253
  "156": {
1254
- "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
1255
  "lstrip": false,
1256
  "normalized": false,
1257
  "rstrip": false,
@@ -1259,7 +1259,7 @@
1259
  "special": false
1260
  },
1261
  "157": {
1262
- "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
1263
  "lstrip": false,
1264
  "normalized": false,
1265
  "rstrip": false,
@@ -1267,7 +1267,7 @@
1267
  "special": false
1268
  },
1269
  "158": {
1270
- "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
1271
  "lstrip": false,
1272
  "normalized": false,
1273
  "rstrip": false,
@@ -1275,7 +1275,7 @@
1275
  "special": false
1276
  },
1277
  "159": {
1278
- "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
1279
  "lstrip": false,
1280
  "normalized": false,
1281
  "rstrip": false,
@@ -1283,7 +1283,7 @@
1283
  "special": false
1284
  },
1285
  "160": {
1286
- "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
1287
  "lstrip": false,
1288
  "normalized": false,
1289
  "rstrip": false,
@@ -1291,7 +1291,7 @@
1291
  "special": false
1292
  },
1293
  "161": {
1294
- "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
1295
  "lstrip": false,
1296
  "normalized": false,
1297
  "rstrip": false,
@@ -1299,7 +1299,7 @@
1299
  "special": false
1300
  },
1301
  "162": {
1302
- "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
1303
  "lstrip": false,
1304
  "normalized": false,
1305
  "rstrip": false,
@@ -1307,7 +1307,7 @@
1307
  "special": false
1308
  },
1309
  "163": {
1310
- "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
1311
  "lstrip": false,
1312
  "normalized": false,
1313
  "rstrip": false,
@@ -1315,7 +1315,7 @@
1315
  "special": false
1316
  },
1317
  "164": {
1318
- "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
1319
  "lstrip": false,
1320
  "normalized": false,
1321
  "rstrip": false,
@@ -1323,7 +1323,7 @@
1323
  "special": false
1324
  },
1325
  "165": {
1326
- "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
1327
  "lstrip": false,
1328
  "normalized": false,
1329
  "rstrip": false,
@@ -1331,7 +1331,7 @@
1331
  "special": false
1332
  },
1333
  "166": {
1334
- "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
1335
  "lstrip": false,
1336
  "normalized": false,
1337
  "rstrip": false,
@@ -1339,7 +1339,7 @@
1339
  "special": false
1340
  },
1341
  "167": {
1342
- "content": "▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁",
1343
  "lstrip": false,
1344
  "normalized": false,
1345
  "rstrip": false,
@@ -53367,7 +53367,6 @@
53367
  "boa_token": "<start_of_audio>",
53368
  "boi_token": "<start_of_image>",
53369
  "bos_token": "<bos>",
53370
- "chat_template": "{{ bos_token }}\n{%- if messages[0]['role'] == 'system' -%}\n {%- if messages[0]['content'] is string -%}\n {%- set first_user_prefix = messages[0]['content'] + '\n\n' -%}\n {%- else -%}\n {%- set first_user_prefix = messages[0]['content'][0]['text'] + '\n\n' -%}\n {%- endif -%}\n {%- set loop_messages = messages[1:] -%}\n{%- else -%}\n {%- set first_user_prefix = \"\" -%}\n {%- set loop_messages = messages -%}\n{%- endif -%}\n{%- for message in loop_messages -%}\n {%- if (message['role'] == 'user') != (loop.index0 % 2 == 0) -%}\n {{ raise_exception(\"Conversation roles must alternate user/assistant/user/assistant/...\") }}\n {%- endif -%}\n {%- if (message['role'] == 'assistant') -%}\n {%- set role = \"model\" -%}\n {%- else -%}\n {%- set role = message['role'] -%}\n {%- endif -%}\n {{ '<start_of_turn>' + role + '\n' + (first_user_prefix if loop.first else \"\") }}\n {%- if message['content'] is string -%}\n {{ message['content'] | trim }}\n {%- elif message['content'] is iterable -%}\n {%- for item in message['content'] -%}\n {%- if item['type'] == 'audio' -%}\n {{ '<audio_soft_token>' }}\n {%- elif item['type'] == 'image' -%}\n {{ '<image_soft_token>' }}\n {%- elif item['type'] == 'text' -%}\n {{ item['text'] | trim }}\n {%- endif -%}\n {%- endfor -%}\n {%- else -%}\n {{ raise_exception(\"Invalid content type\") }}\n {%- endif -%}\n {{ '<end_of_turn>\n' }}\n{%- endfor -%}\n{%- if add_generation_prompt -%}\n {{'<start_of_turn>model\n'}}\n{%- endif -%}\n",
53371
  "clean_up_tokenization_spaces": false,
53372
  "eoa_token": "<end_of_audio>",
53373
  "eoi_token": "<end_of_image>",
@@ -53388,5 +53387,6 @@
53388
  "spaces_between_special_tokens": false,
53389
  "tokenizer_class": "GemmaTokenizer",
53390
  "unk_token": "<unk>",
53391
- "use_default_system_prompt": false
53392
- }
 
 
1107
  "special": false
1108
  },
1109
  "138": {
1110
+ "content": "\u2581\u2581",
1111
  "lstrip": false,
1112
  "normalized": false,
1113
  "rstrip": false,
 
1115
  "special": false
1116
  },
1117
  "139": {
1118
+ "content": "\u2581\u2581\u2581",
1119
  "lstrip": false,
1120
  "normalized": false,
1121
  "rstrip": false,
 
1123
  "special": false
1124
  },
1125
  "140": {
1126
+ "content": "\u2581\u2581\u2581\u2581",
1127
  "lstrip": false,
1128
  "normalized": false,
1129
  "rstrip": false,
 
1131
  "special": false
1132
  },
1133
  "141": {
1134
+ "content": "\u2581\u2581\u2581\u2581\u2581",
1135
  "lstrip": false,
1136
  "normalized": false,
1137
  "rstrip": false,
 
1139
  "special": false
1140
  },
1141
  "142": {
1142
+ "content": "\u2581\u2581\u2581\u2581\u2581\u2581",
1143
  "lstrip": false,
1144
  "normalized": false,
1145
  "rstrip": false,
 
1147
  "special": false
1148
  },
1149
  "143": {
1150
+ "content": "\u2581\u2581\u2581\u2581\u2581\u2581\u2581",
1151
  "lstrip": false,
1152
  "normalized": false,
1153
  "rstrip": false,
 
1155
  "special": false
1156
  },
1157
  "144": {
1158
+ "content": "\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581",
1159
  "lstrip": false,
1160
  "normalized": false,
1161
  "rstrip": false,
 
1163
  "special": false
1164
  },
1165
  "145": {
1166
+ "content": "\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581",
1167
  "lstrip": false,
1168
  "normalized": false,
1169
  "rstrip": false,
 
1171
  "special": false
1172
  },
1173
  "146": {
1174
+ "content": "\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581",
1175
  "lstrip": false,
1176
  "normalized": false,
1177
  "rstrip": false,
 
1179
  "special": false
1180
  },
1181
  "147": {
1182
+ "content": "\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581",
1183
  "lstrip": false,
1184
  "normalized": false,
1185
  "rstrip": false,
 
1187
  "special": false
1188
  },
1189
  "148": {
1190
+ "content": "\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581",
1191
  "lstrip": false,
1192
  "normalized": false,
1193
  "rstrip": false,
 
1195
  "special": false
1196
  },
1197
  "149": {
1198
+ "content": "\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581",
1199
  "lstrip": false,
1200
  "normalized": false,
1201
  "rstrip": false,
 
1203
  "special": false
1204
  },
1205
  "150": {
1206
+ "content": "\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581",
1207
  "lstrip": false,
1208
  "normalized": false,
1209
  "rstrip": false,
 
1211
  "special": false
1212
  },
1213
  "151": {
1214
+ "content": "\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581",
1215
  "lstrip": false,
1216
  "normalized": false,
1217
  "rstrip": false,
 
1219
  "special": false
1220
  },
1221
  "152": {
1222
+ "content": "\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581",
1223
  "lstrip": false,
1224
  "normalized": false,
1225
  "rstrip": false,
 
1227
  "special": false
1228
  },
1229
  "153": {
1230
+ "content": "\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581",
1231
  "lstrip": false,
1232
  "normalized": false,
1233
  "rstrip": false,
 
1235
  "special": false
1236
  },
1237
  "154": {
1238
+ "content": "\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581",
1239
  "lstrip": false,
1240
  "normalized": false,
1241
  "rstrip": false,
 
1243
  "special": false
1244
  },
1245
  "155": {
1246
+ "content": "\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581",
1247
  "lstrip": false,
1248
  "normalized": false,
1249
  "rstrip": false,
 
1251
  "special": false
1252
  },
1253
  "156": {
1254
+ "content": "\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581",
1255
  "lstrip": false,
1256
  "normalized": false,
1257
  "rstrip": false,
 
1259
  "special": false
1260
  },
1261
  "157": {
1262
+ "content": "\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581",
1263
  "lstrip": false,
1264
  "normalized": false,
1265
  "rstrip": false,
 
1267
  "special": false
1268
  },
1269
  "158": {
1270
+ "content": "\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581",
1271
  "lstrip": false,
1272
  "normalized": false,
1273
  "rstrip": false,
 
1275
  "special": false
1276
  },
1277
  "159": {
1278
+ "content": "\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581",
1279
  "lstrip": false,
1280
  "normalized": false,
1281
  "rstrip": false,
 
1283
  "special": false
1284
  },
1285
  "160": {
1286
+ "content": "\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581",
1287
  "lstrip": false,
1288
  "normalized": false,
1289
  "rstrip": false,
 
1291
  "special": false
1292
  },
1293
  "161": {
1294
+ "content": "\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581",
1295
  "lstrip": false,
1296
  "normalized": false,
1297
  "rstrip": false,
 
1299
  "special": false
1300
  },
1301
  "162": {
1302
+ "content": "\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581",
1303
  "lstrip": false,
1304
  "normalized": false,
1305
  "rstrip": false,
 
1307
  "special": false
1308
  },
1309
  "163": {
1310
+ "content": "\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581",
1311
  "lstrip": false,
1312
  "normalized": false,
1313
  "rstrip": false,
 
1315
  "special": false
1316
  },
1317
  "164": {
1318
+ "content": "\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581",
1319
  "lstrip": false,
1320
  "normalized": false,
1321
  "rstrip": false,
 
1323
  "special": false
1324
  },
1325
  "165": {
1326
+ "content": "\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581",
1327
  "lstrip": false,
1328
  "normalized": false,
1329
  "rstrip": false,
 
1331
  "special": false
1332
  },
1333
  "166": {
1334
+ "content": "\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581",
1335
  "lstrip": false,
1336
  "normalized": false,
1337
  "rstrip": false,
 
1339
  "special": false
1340
  },
1341
  "167": {
1342
+ "content": "\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581\u2581",
1343
  "lstrip": false,
1344
  "normalized": false,
1345
  "rstrip": false,
 
53367
  "boa_token": "<start_of_audio>",
53368
  "boi_token": "<start_of_image>",
53369
  "bos_token": "<bos>",
 
53370
  "clean_up_tokenization_spaces": false,
53371
  "eoa_token": "<end_of_audio>",
53372
  "eoi_token": "<end_of_image>",
 
53387
  "spaces_between_special_tokens": false,
53388
  "tokenizer_class": "GemmaTokenizer",
53389
  "unk_token": "<unk>",
53390
+ "use_default_system_prompt": false,
53391
+ "chat_template": "{{ bos_token }}\n{%- if messages[0]['role'] == 'system' -%}\n {%- if messages[0]['content'] is string -%}\n {%- set first_user_prefix = messages[0]['content'] + '\n\n' -%}\n {%- else -%}\n {%- set first_user_prefix = messages[0]['content'][0]['text'] + '\n\n' -%}\n {%- endif -%}\n {%- set loop_messages = messages[1:] -%}\n{%- else -%}\n {%- set first_user_prefix = \"\" -%}\n {%- set loop_messages = messages -%}\n{%- endif -%}\n{%- for message in loop_messages -%}\n {%- if (message['role'] == 'user') != (loop.index0 % 2 == 0) -%}\n {{ raise_exception(\"Conversation roles must alternate user/assistant/user/assistant/...\") }}\n {%- endif -%}\n {%- if (message['role'] == 'assistant') -%}\n {%- set role = \"model\" -%}\n {%- else -%}\n {%- set role = message['role'] -%}\n {%- endif -%}\n {{ '<start_of_turn>' + role + '\n' + (first_user_prefix if loop.first else \"\") }}\n {%- if message['content'] is string -%}\n {{ message['content'] | trim }}\n {%- elif message['content'] is iterable -%}\n {%- for item in message['content'] -%}\n {%- if item['type'] == 'audio' -%}\n {{ '<audio_soft_token>' }}\n {%- elif item['type'] == 'image' -%}\n {{ '<image_soft_token>' }}\n {%- elif item['type'] == 'text' -%}\n {{ item['text'] | trim }}\n {%- endif -%}\n {%- endfor -%}\n {%- else -%}\n {{ raise_exception(\"Invalid content type\") }}\n {%- endif -%}\n {{ '<end_of_turn>\n' }}\n{%- endfor -%}\n{%- if add_generation_prompt -%}\n {{'<start_of_turn>model\n'}}\n{%- endif -%}\n"
53392
+ }