Update config.json
Browse files- config.json +101 -9
config.json
CHANGED
|
@@ -1,4 +1,12 @@
|
|
| 1 |
{
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2 |
"architectures": [
|
| 3 |
"IsaacForConditionalGeneration"
|
| 4 |
],
|
|
@@ -53,20 +61,92 @@
|
|
| 53 |
"num_attention_heads": 16,
|
| 54 |
"num_hidden_layers": 28,
|
| 55 |
"num_key_value_heads": 8,
|
|
|
|
| 56 |
"pixel_shuffle_scale": 2,
|
| 57 |
"rms_norm_eps": 1e-06,
|
| 58 |
-
"
|
| 59 |
-
"mrope_interleaved": true,
|
| 60 |
-
"mrope_section": null,
|
| 61 |
-
"rope_type": "default"
|
| 62 |
-
},
|
| 63 |
-
"rope_theta": 1000000.0,
|
| 64 |
"sliding_window": null,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 65 |
"tie_word_embeddings": false,
|
| 66 |
-
"transformers_version": "
|
| 67 |
"use_cache": true,
|
| 68 |
"use_sliding_window": false,
|
| 69 |
-
"video_patch_size": 16,
|
| 70 |
"vision_config": {
|
| 71 |
"attention_dropout": 0.0,
|
| 72 |
"hidden_act": "gelu_pytorch_tanh",
|
|
@@ -74,7 +154,7 @@
|
|
| 74 |
"image_size": 256,
|
| 75 |
"intermediate_size": 4304,
|
| 76 |
"layer_norm_eps": 1e-06,
|
| 77 |
-
"model_type": "
|
| 78 |
"num_attention_heads": 16,
|
| 79 |
"num_channels": 3,
|
| 80 |
"num_hidden_layers": 27,
|
|
@@ -83,7 +163,19 @@
|
|
| 83 |
"pixel_shuffle_scale_factor": 2
|
| 84 |
},
|
| 85 |
"vision_max_num_patches": 6144,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 86 |
"vision_min_num_patches": 256,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 87 |
"vision_token": "<image>",
|
| 88 |
"vocab_size": 151936
|
| 89 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"_rope_parameters": {
|
| 3 |
+
"rope_theta": 1000000,
|
| 4 |
+
"rope_type": "default"
|
| 5 |
+
},
|
| 6 |
+
"_rope_scaling": {
|
| 7 |
+
"rope_theta": 1000000,
|
| 8 |
+
"rope_type": "default"
|
| 9 |
+
},
|
| 10 |
"architectures": [
|
| 11 |
"IsaacForConditionalGeneration"
|
| 12 |
],
|
|
|
|
| 61 |
"num_attention_heads": 16,
|
| 62 |
"num_hidden_layers": 28,
|
| 63 |
"num_key_value_heads": 8,
|
| 64 |
+
"pad_token_id": null,
|
| 65 |
"pixel_shuffle_scale": 2,
|
| 66 |
"rms_norm_eps": 1e-06,
|
| 67 |
+
"rope_theta": 1000000,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 68 |
"sliding_window": null,
|
| 69 |
+
"text_config": {
|
| 70 |
+
"_name_or_path": "hf-checkpoint",
|
| 71 |
+
"architectures": [
|
| 72 |
+
"IsaacForConditionalGeneration"
|
| 73 |
+
],
|
| 74 |
+
"attention_bias": false,
|
| 75 |
+
"attention_dropout": 0.0,
|
| 76 |
+
"bos_token_id": 151643,
|
| 77 |
+
"dtype": "float32",
|
| 78 |
+
"eos_token_id": 151645,
|
| 79 |
+
"head_dim": 128,
|
| 80 |
+
"hidden_act": "silu",
|
| 81 |
+
"hidden_size": 2048,
|
| 82 |
+
"initializer_range": 0.02,
|
| 83 |
+
"intermediate_size": 6144,
|
| 84 |
+
"layer_types": [
|
| 85 |
+
"full_attention",
|
| 86 |
+
"full_attention",
|
| 87 |
+
"full_attention",
|
| 88 |
+
"full_attention",
|
| 89 |
+
"full_attention",
|
| 90 |
+
"full_attention",
|
| 91 |
+
"full_attention",
|
| 92 |
+
"full_attention",
|
| 93 |
+
"full_attention",
|
| 94 |
+
"full_attention",
|
| 95 |
+
"full_attention",
|
| 96 |
+
"full_attention",
|
| 97 |
+
"full_attention",
|
| 98 |
+
"full_attention",
|
| 99 |
+
"full_attention",
|
| 100 |
+
"full_attention",
|
| 101 |
+
"full_attention",
|
| 102 |
+
"full_attention",
|
| 103 |
+
"full_attention",
|
| 104 |
+
"full_attention",
|
| 105 |
+
"full_attention",
|
| 106 |
+
"full_attention",
|
| 107 |
+
"full_attention",
|
| 108 |
+
"full_attention",
|
| 109 |
+
"full_attention",
|
| 110 |
+
"full_attention",
|
| 111 |
+
"full_attention",
|
| 112 |
+
"full_attention"
|
| 113 |
+
],
|
| 114 |
+
"max_position_embeddings": 40960,
|
| 115 |
+
"max_window_layers": 28,
|
| 116 |
+
"model_type": "qwen3",
|
| 117 |
+
"num_attention_heads": 16,
|
| 118 |
+
"num_hidden_layers": 28,
|
| 119 |
+
"num_key_value_heads": 8,
|
| 120 |
+
"pad_token_id": null,
|
| 121 |
+
"pixel_shuffle_scale": 2,
|
| 122 |
+
"rms_norm_eps": 1e-06,
|
| 123 |
+
"rope_parameters": {
|
| 124 |
+
"rope_theta": 1000000,
|
| 125 |
+
"rope_type": "default"
|
| 126 |
+
},
|
| 127 |
+
"sliding_window": null,
|
| 128 |
+
"tie_word_embeddings": false,
|
| 129 |
+
"use_cache": true,
|
| 130 |
+
"use_sliding_window": false,
|
| 131 |
+
"vision_max_num_patches": 6144,
|
| 132 |
+
"vision_mean": [
|
| 133 |
+
0.5,
|
| 134 |
+
0.5,
|
| 135 |
+
0.5
|
| 136 |
+
],
|
| 137 |
+
"vision_min_num_patches": 256,
|
| 138 |
+
"vision_patch_size": 16,
|
| 139 |
+
"vision_std": [
|
| 140 |
+
0.5,
|
| 141 |
+
0.5,
|
| 142 |
+
0.5
|
| 143 |
+
],
|
| 144 |
+
"vocab_size": 151936
|
| 145 |
+
},
|
| 146 |
"tie_word_embeddings": false,
|
| 147 |
+
"transformers_version": "5.0.0",
|
| 148 |
"use_cache": true,
|
| 149 |
"use_sliding_window": false,
|
|
|
|
| 150 |
"vision_config": {
|
| 151 |
"attention_dropout": 0.0,
|
| 152 |
"hidden_act": "gelu_pytorch_tanh",
|
|
|
|
| 154 |
"image_size": 256,
|
| 155 |
"intermediate_size": 4304,
|
| 156 |
"layer_norm_eps": 1e-06,
|
| 157 |
+
"model_type": "isaac_vision",
|
| 158 |
"num_attention_heads": 16,
|
| 159 |
"num_channels": 3,
|
| 160 |
"num_hidden_layers": 27,
|
|
|
|
| 163 |
"pixel_shuffle_scale_factor": 2
|
| 164 |
},
|
| 165 |
"vision_max_num_patches": 6144,
|
| 166 |
+
"vision_mean": [
|
| 167 |
+
0.5,
|
| 168 |
+
0.5,
|
| 169 |
+
0.5
|
| 170 |
+
],
|
| 171 |
"vision_min_num_patches": 256,
|
| 172 |
+
"vision_patch_size": 16,
|
| 173 |
+
"vision_rescale_factor": 0.00392156862745098,
|
| 174 |
+
"vision_std": [
|
| 175 |
+
0.5,
|
| 176 |
+
0.5,
|
| 177 |
+
0.5
|
| 178 |
+
],
|
| 179 |
"vision_token": "<image>",
|
| 180 |
"vocab_size": 151936
|
| 181 |
}
|