Upload config.json with huggingface_hub
Browse files- config.json +204 -0
config.json
ADDED
|
@@ -0,0 +1,204 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "/cm/archive/namnv78_new/revise_checkpoints/Xphi35-siglip224/pft",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"LlavaPhiForCausalLM"
|
| 5 |
+
],
|
| 6 |
+
"attention_bias": false,
|
| 7 |
+
"attention_dropout": 0.0,
|
| 8 |
+
"auto_map": {
|
| 9 |
+
"AutoConfig": "configuration_phi3.Phi3Config",
|
| 10 |
+
"AutoModelForCausalLM": "modeling_phi3.Phi3ForCausalLM"
|
| 11 |
+
},
|
| 12 |
+
"bal_comp_loss_coef": 0.01,
|
| 13 |
+
"balance_loss_coef": 0.01,
|
| 14 |
+
"bos_token_id": 1,
|
| 15 |
+
"clip_smoe": true,
|
| 16 |
+
"diversity_loss_coef": 0.01,
|
| 17 |
+
"dropout": false,
|
| 18 |
+
"e_loss_coef": 0.001,
|
| 19 |
+
"embd_pdrop": 0.0,
|
| 20 |
+
"entropy_advance_loss": false,
|
| 21 |
+
"eos_token_id": 32000,
|
| 22 |
+
"freeze_backbone": false,
|
| 23 |
+
"freeze_mm_mlp_adapter": false,
|
| 24 |
+
"hidden_act": "silu",
|
| 25 |
+
"hidden_size": 3072,
|
| 26 |
+
"hybrid": false,
|
| 27 |
+
"image_aspect_ratio": "pad",
|
| 28 |
+
"init_weight": true,
|
| 29 |
+
"initializer_range": 0.02,
|
| 30 |
+
"intermediate_size": 8192,
|
| 31 |
+
"is_cosine": false,
|
| 32 |
+
"is_norm_weight": false,
|
| 33 |
+
"local_rank": 0,
|
| 34 |
+
"loss1": "balanceloss",
|
| 35 |
+
"loss2": "zloss",
|
| 36 |
+
"luna": false,
|
| 37 |
+
"max_compete_in_iter": 3,
|
| 38 |
+
"max_position_embeddings": 131072,
|
| 39 |
+
"mlp_norm": false,
|
| 40 |
+
"mlp_smoe": true,
|
| 41 |
+
"mm_hidden_size": 1152,
|
| 42 |
+
"mm_patch_merge_type": "flat",
|
| 43 |
+
"mm_projector_lr": null,
|
| 44 |
+
"mm_projector_type": "moe",
|
| 45 |
+
"mm_use_im_patch_token": false,
|
| 46 |
+
"mm_use_im_start_end": false,
|
| 47 |
+
"mm_vision_select_feature": "patch",
|
| 48 |
+
"mm_vision_select_layer": -2,
|
| 49 |
+
"mm_vision_tower": "google/siglip-so400m-patch14-224",
|
| 50 |
+
"model_name_or_path": "/cm/archive/namnv78_new/revise_checkpoints/Xphi35-siglip224/pft",
|
| 51 |
+
"model_type": "llava_phi",
|
| 52 |
+
"moe_name": "smoe_share",
|
| 53 |
+
"moe_relu_l1_reg_coeff_multiplier": 1.2,
|
| 54 |
+
"mp_pixel_shuffle_factor": 1,
|
| 55 |
+
"norm_softmax": false,
|
| 56 |
+
"normalization": true,
|
| 57 |
+
"num_attention_heads": 32,
|
| 58 |
+
"num_experts": 6,
|
| 59 |
+
"num_hidden_layers": 32,
|
| 60 |
+
"num_key_value_heads": 32,
|
| 61 |
+
"num_layers": 3,
|
| 62 |
+
"num_selected": 3,
|
| 63 |
+
"number_of_previous_tokens": 2,
|
| 64 |
+
"original_max_position_embeddings": 4096,
|
| 65 |
+
"pad_token_id": 32000,
|
| 66 |
+
"pretrain_mm_mlp_adapter": "/cm/archive/namnv78_new/revise_checkpoints/Xphi35-siglip224/pft/mm_projector.bin",
|
| 67 |
+
"rate_compete": 0.2,
|
| 68 |
+
"rate_flip": 0.05,
|
| 69 |
+
"resid_pdrop": 0.0,
|
| 70 |
+
"rms_norm_eps": 1e-05,
|
| 71 |
+
"rope_scaling": {
|
| 72 |
+
"long_factor": [
|
| 73 |
+
1.0800000429153442,
|
| 74 |
+
1.1100000143051147,
|
| 75 |
+
1.1399999856948853,
|
| 76 |
+
1.340000033378601,
|
| 77 |
+
1.5899999141693115,
|
| 78 |
+
1.600000023841858,
|
| 79 |
+
1.6200000047683716,
|
| 80 |
+
2.620000123977661,
|
| 81 |
+
3.2300000190734863,
|
| 82 |
+
3.2300000190734863,
|
| 83 |
+
4.789999961853027,
|
| 84 |
+
7.400000095367432,
|
| 85 |
+
7.700000286102295,
|
| 86 |
+
9.09000015258789,
|
| 87 |
+
12.199999809265137,
|
| 88 |
+
17.670000076293945,
|
| 89 |
+
24.46000099182129,
|
| 90 |
+
28.57000160217285,
|
| 91 |
+
30.420001983642578,
|
| 92 |
+
30.840002059936523,
|
| 93 |
+
32.590003967285156,
|
| 94 |
+
32.93000411987305,
|
| 95 |
+
42.320003509521484,
|
| 96 |
+
44.96000289916992,
|
| 97 |
+
50.340003967285156,
|
| 98 |
+
50.45000457763672,
|
| 99 |
+
57.55000305175781,
|
| 100 |
+
57.93000411987305,
|
| 101 |
+
58.21000289916992,
|
| 102 |
+
60.1400032043457,
|
| 103 |
+
62.61000442504883,
|
| 104 |
+
62.62000274658203,
|
| 105 |
+
62.71000289916992,
|
| 106 |
+
63.1400032043457,
|
| 107 |
+
63.1400032043457,
|
| 108 |
+
63.77000427246094,
|
| 109 |
+
63.93000411987305,
|
| 110 |
+
63.96000289916992,
|
| 111 |
+
63.970001220703125,
|
| 112 |
+
64.02999877929688,
|
| 113 |
+
64.06999969482422,
|
| 114 |
+
64.08000183105469,
|
| 115 |
+
64.12000274658203,
|
| 116 |
+
64.41000366210938,
|
| 117 |
+
64.4800033569336,
|
| 118 |
+
64.51000213623047,
|
| 119 |
+
64.52999877929688,
|
| 120 |
+
64.83999633789062
|
| 121 |
+
],
|
| 122 |
+
"short_factor": [
|
| 123 |
+
1.0,
|
| 124 |
+
1.0199999809265137,
|
| 125 |
+
1.0299999713897705,
|
| 126 |
+
1.0299999713897705,
|
| 127 |
+
1.0499999523162842,
|
| 128 |
+
1.0499999523162842,
|
| 129 |
+
1.0499999523162842,
|
| 130 |
+
1.0499999523162842,
|
| 131 |
+
1.0499999523162842,
|
| 132 |
+
1.0699999332427979,
|
| 133 |
+
1.0999999046325684,
|
| 134 |
+
1.1099998950958252,
|
| 135 |
+
1.1599998474121094,
|
| 136 |
+
1.1599998474121094,
|
| 137 |
+
1.1699998378753662,
|
| 138 |
+
1.2899998426437378,
|
| 139 |
+
1.339999794960022,
|
| 140 |
+
1.679999828338623,
|
| 141 |
+
1.7899998426437378,
|
| 142 |
+
1.8199998140335083,
|
| 143 |
+
1.8499997854232788,
|
| 144 |
+
1.8799997568130493,
|
| 145 |
+
1.9099997282028198,
|
| 146 |
+
1.9399996995925903,
|
| 147 |
+
1.9899996519088745,
|
| 148 |
+
2.0199997425079346,
|
| 149 |
+
2.0199997425079346,
|
| 150 |
+
2.0199997425079346,
|
| 151 |
+
2.0199997425079346,
|
| 152 |
+
2.0199997425079346,
|
| 153 |
+
2.0199997425079346,
|
| 154 |
+
2.0299997329711914,
|
| 155 |
+
2.0299997329711914,
|
| 156 |
+
2.0299997329711914,
|
| 157 |
+
2.0299997329711914,
|
| 158 |
+
2.0299997329711914,
|
| 159 |
+
2.0299997329711914,
|
| 160 |
+
2.0299997329711914,
|
| 161 |
+
2.0299997329711914,
|
| 162 |
+
2.0299997329711914,
|
| 163 |
+
2.0799996852874756,
|
| 164 |
+
2.0899996757507324,
|
| 165 |
+
2.189999580383301,
|
| 166 |
+
2.2199995517730713,
|
| 167 |
+
2.5899994373321533,
|
| 168 |
+
2.729999542236328,
|
| 169 |
+
2.749999523162842,
|
| 170 |
+
2.8399994373321533
|
| 171 |
+
],
|
| 172 |
+
"type": "longrope"
|
| 173 |
+
},
|
| 174 |
+
"rope_theta": 10000.0,
|
| 175 |
+
"router_loss_coef": 0.01,
|
| 176 |
+
"router_theta": 0.1,
|
| 177 |
+
"router_z_loss_coef": 0.001,
|
| 178 |
+
"scales": [
|
| 179 |
+
1,
|
| 180 |
+
3
|
| 181 |
+
],
|
| 182 |
+
"sliding_window": 262144,
|
| 183 |
+
"sparse_upcycling": true,
|
| 184 |
+
"std_gate": 0.02,
|
| 185 |
+
"strategy_train": "base",
|
| 186 |
+
"tie_word_embeddings": false,
|
| 187 |
+
"tokenizer_model_max_length": 2048,
|
| 188 |
+
"tokenizer_padding_side": "right",
|
| 189 |
+
"topk_max": 2,
|
| 190 |
+
"topk_min": 1,
|
| 191 |
+
"torch_dtype": "bfloat16",
|
| 192 |
+
"training": true,
|
| 193 |
+
"transformers_version": "4.43.0",
|
| 194 |
+
"tune_mm_mlp_adapter": false,
|
| 195 |
+
"unit_test": true,
|
| 196 |
+
"use_cache": true,
|
| 197 |
+
"use_mm_proj": true,
|
| 198 |
+
"use_old": false,
|
| 199 |
+
"version": "phi35",
|
| 200 |
+
"vision_tower": "google/siglip-so400m-patch14-224",
|
| 201 |
+
"vision_tower_dir": "/cm/archive/namnv78_new/revise_checkpoints/Xphi35-siglip224/pft/clip.bin",
|
| 202 |
+
"vocab_size": 32064,
|
| 203 |
+
"warm_up": 0.05
|
| 204 |
+
}
|