| { |
| "architecture_type": "uni-encoder", |
| "architectures": [ |
| "GLiClassModel" |
| ], |
| "class_token_index": 250100, |
| "class_token_pooling": "average", |
| "contrastive_loss_coef": 0.0, |
| "dropout": 0.3, |
| "dtype": "bfloat16", |
| "embed_class_token": true, |
| "encoder_config": { |
| "_name_or_path": "google/mt5-xl", |
| "architectures": [ |
| "MT5ForConditionalGeneration" |
| ], |
| "attention_type": "triton-basic", |
| "classifier_dropout": 0.0, |
| "d_ff": 5120, |
| "d_kv": 64, |
| "d_model": 2048, |
| "dense_act_fn": "gelu_new", |
| "dropout_rate": 0.1, |
| "feed_forward_proj": "gated-gelu", |
| "global_block_size": 32, |
| "initializer_factor": 1.0, |
| "is_encoder_decoder": false, |
| "is_gated_act": true, |
| "layer_norm_epsilon": 1e-06, |
| "model_type": "mt5", |
| "num_decoder_layers": 24, |
| "num_heads": 32, |
| "num_layers": 24, |
| "output_past": true, |
| "relative_attention_max_distance": 128, |
| "relative_attention_num_buckets": 32, |
| "sliding_window": 512, |
| "use_cache": false, |
| "use_triton": true, |
| "vocab_size": 250103 |
| }, |
| "encoder_layer_id": -1, |
| "encoder_model_name": "google/mt5-xl", |
| "eos_token_id": 1, |
| "example_token_index": 250102, |
| "extract_text_features": true, |
| "focal_loss_alpha": -1, |
| "focal_loss_gamma": -1, |
| "focal_loss_reduction": "none", |
| "hidden_size": 2048, |
| "ignore_index": -100, |
| "initializer_range": 0.03, |
| "label_model_config": null, |
| "label_model_name": null, |
| "layer_wise": false, |
| "logit_scale_init_value": 2.6592, |
| "max_num_classes": 256, |
| "model_type": "GLiClass", |
| "normalize_features": false, |
| "pad_token_id": 0, |
| "pooling_strategy": "pass", |
| "projector_hidden_act": "gelu", |
| "prompt_first": true, |
| "scorer_attn_dropout": 0.1, |
| "scorer_mlp_hidden_size": 1024, |
| "scorer_num_heads": 16, |
| "scorer_type": "cross-attn", |
| "scorer_use_sequence_packing": true, |
| "shuffle_labels": true, |
| "squeeze_layers": false, |
| "text_token_index": 250101, |
| "transformers_version": "4.57.6", |
| "use_lstm": false, |
| "use_segment_embeddings": false, |
| "vocab_size": 250103 |
| } |
|
|
