Raise per-image vision soft-token budget from 280 to 1120

#41
Files changed (2) hide show
  1. config.json +2 -2
  2. processor_config.json +3 -3
config.json CHANGED
@@ -149,7 +149,7 @@
149
  "attention_bias": false,
150
  "attention_dropout": 0.0,
151
  "chunk_size_feed_forward": 0,
152
- "default_output_length": 280,
153
  "dtype": "bfloat16",
154
  "global_head_dim": 64,
155
  "head_dim": 64,
@@ -186,5 +186,5 @@
186
  "standardize": false,
187
  "use_clipped_linears": true
188
  },
189
- "vision_soft_tokens_per_image": 280
190
  }
 
149
  "attention_bias": false,
150
  "attention_dropout": 0.0,
151
  "chunk_size_feed_forward": 0,
152
+ "default_output_length": 1120,
153
  "dtype": "bfloat16",
154
  "global_head_dim": 64,
155
  "head_dim": 64,
 
186
  "standardize": false,
187
  "use_clipped_linears": true
188
  },
189
+ "vision_soft_tokens_per_image": 1120
190
  }
processor_config.json CHANGED
@@ -33,19 +33,19 @@
33
  0.0
34
  ],
35
  "image_processor_type": "Gemma4ImageProcessor",
36
- "image_seq_length": 280,
37
  "image_std": [
38
  1.0,
39
  1.0,
40
  1.0
41
  ],
42
- "max_soft_tokens": 280,
43
  "patch_size": 16,
44
  "pooling_kernel_size": 3,
45
  "resample": 3,
46
  "rescale_factor": 0.00392156862745098
47
  },
48
- "image_seq_length": 280,
49
  "processor_class": "Gemma4Processor",
50
  "video_processor": {
51
  "do_convert_rgb": true,
 
33
  0.0
34
  ],
35
  "image_processor_type": "Gemma4ImageProcessor",
36
+ "image_seq_length": 1120,
37
  "image_std": [
38
  1.0,
39
  1.0,
40
  1.0
41
  ],
42
+ "max_soft_tokens": 1120,
43
  "patch_size": 16,
44
  "pooling_kernel_size": 3,
45
  "resample": 3,
46
  "rescale_factor": 0.00392156862745098
47
  },
48
+ "image_seq_length": 1120,
49
  "processor_class": "Gemma4Processor",
50
  "video_processor": {
51
  "do_convert_rgb": true,