Upload config.json with huggingface_hub
Browse files- config.json +82 -0
config.json
ADDED
|
@@ -0,0 +1,82 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "roberta-base",
|
| 3 |
+
"adapters": {
|
| 4 |
+
"adapters": {
|
| 5 |
+
"super_glue_multirc": "9076f36a74755ac4"
|
| 6 |
+
},
|
| 7 |
+
"config_map": {
|
| 8 |
+
"9076f36a74755ac4": {
|
| 9 |
+
"adapter_residual_before_ln": false,
|
| 10 |
+
"cross_adapter": false,
|
| 11 |
+
"factorized_phm_W": true,
|
| 12 |
+
"factorized_phm_rule": false,
|
| 13 |
+
"hypercomplex_nonlinearity": "glorot-uniform",
|
| 14 |
+
"init_weights": "bert",
|
| 15 |
+
"inv_adapter": null,
|
| 16 |
+
"inv_adapter_reduction_factor": null,
|
| 17 |
+
"is_parallel": false,
|
| 18 |
+
"learn_phm": true,
|
| 19 |
+
"leave_out": [],
|
| 20 |
+
"ln_after": false,
|
| 21 |
+
"ln_before": false,
|
| 22 |
+
"mh_adapter": false,
|
| 23 |
+
"non_linearity": "relu",
|
| 24 |
+
"original_ln_after": true,
|
| 25 |
+
"original_ln_before": true,
|
| 26 |
+
"output_adapter": true,
|
| 27 |
+
"phm_bias": true,
|
| 28 |
+
"phm_c_init": "normal",
|
| 29 |
+
"phm_dim": 4,
|
| 30 |
+
"phm_init_range": 0.0001,
|
| 31 |
+
"phm_layer": false,
|
| 32 |
+
"phm_rank": 1,
|
| 33 |
+
"reduction_factor": 16,
|
| 34 |
+
"residual_before_ln": true,
|
| 35 |
+
"scaling": 1.0,
|
| 36 |
+
"shared_W_phm": false,
|
| 37 |
+
"shared_phm_rule": true,
|
| 38 |
+
"use_gating": false
|
| 39 |
+
}
|
| 40 |
+
},
|
| 41 |
+
"fusion_config_map": {},
|
| 42 |
+
"fusions": {}
|
| 43 |
+
},
|
| 44 |
+
"architectures": [
|
| 45 |
+
"RobertaModelWithHeads"
|
| 46 |
+
],
|
| 47 |
+
"attention_probs_dropout_prob": 0.1,
|
| 48 |
+
"bos_token_id": 0,
|
| 49 |
+
"classifier_dropout": null,
|
| 50 |
+
"eos_token_id": 2,
|
| 51 |
+
"hidden_act": "gelu",
|
| 52 |
+
"hidden_dropout_prob": 0.1,
|
| 53 |
+
"hidden_size": 768,
|
| 54 |
+
"initializer_range": 0.02,
|
| 55 |
+
"intermediate_size": 3072,
|
| 56 |
+
"layer_norm_eps": 1e-05,
|
| 57 |
+
"max_position_embeddings": 514,
|
| 58 |
+
"model_type": "roberta",
|
| 59 |
+
"num_attention_heads": 12,
|
| 60 |
+
"num_hidden_layers": 12,
|
| 61 |
+
"pad_token_id": 1,
|
| 62 |
+
"position_embedding_type": "absolute",
|
| 63 |
+
"prediction_heads": {
|
| 64 |
+
"super_glue_multirc": {
|
| 65 |
+
"activation_function": "tanh",
|
| 66 |
+
"bias": true,
|
| 67 |
+
"head_type": "classification",
|
| 68 |
+
"label2id": {
|
| 69 |
+
"False": 0,
|
| 70 |
+
"True": 1
|
| 71 |
+
},
|
| 72 |
+
"layers": 2,
|
| 73 |
+
"num_labels": 2,
|
| 74 |
+
"use_pooler": false
|
| 75 |
+
}
|
| 76 |
+
},
|
| 77 |
+
"torch_dtype": "float32",
|
| 78 |
+
"transformers_version": "4.21.3",
|
| 79 |
+
"type_vocab_size": 1,
|
| 80 |
+
"use_cache": false,
|
| 81 |
+
"vocab_size": 50265
|
| 82 |
+
}
|