aieng-lab/codet5p-220m_comment-type-java

1{2  "_name_or_path": "Salesforce/codet5p-220m",3  "architectures": [4    "T5ForSequenceClassification"5  ],6  "bos_token_id": 1,7  "classifier_dropout": 0.0,8  "d_ff": 3072,9  "d_kv": 64,10  "d_model": 768,11  "decoder_start_token_id": 0,12  "dense_act_fn": "relu",13  "dropout_rate": 0.1,14  "eos_token_id": 2,15  "feed_forward_proj": "relu",16  "id2label": {17    "0": "LABEL_0",18    "1": "LABEL_1",19    "2": "LABEL_2",20    "3": "LABEL_3",21    "4": "LABEL_4",22    "5": "LABEL_5",23    "6": "LABEL_6"24  },25  "initializer_factor": 1.0,26  "is_encoder_decoder": true,27  "is_gated_act": false,28  "label2id": {29    "LABEL_0": 0,30    "LABEL_1": 1,31    "LABEL_2": 2,32    "LABEL_3": 3,33    "LABEL_4": 4,34    "LABEL_5": 5,35    "LABEL_6": 636  },37  "layer_norm_epsilon": 1e-06,38  "model_type": "t5",39  "n_positions": 512,40  "num_decoder_layers": 12,41  "num_heads": 12,42  "num_layers": 12,43  "output_past": true,44  "pad_token_id": 0,45  "problem_type": "multi_label_classification",46  "relative_attention_max_distance": 128,47  "relative_attention_num_buckets": 32,48  "torch_dtype": "bfloat16",49  "transformers_version": "4.48.3",50  "use_cache": true,51  "vocab_size": 3210052}53