LonlyDWolf/CodeT5p-220M-For-Code-Generation
07
1{2 "_name_or_path": "Salesforce/codet5p-220m",3 "architectures": [4 "T5ForConditionalGeneration"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 "initializer_factor": 1.0,17 "is_encoder_decoder": true,18 "is_gated_act": false,19 "layer_norm_epsilon": 1e-06,20 "model_type": "t5",21 "n_positions": 512,22 "num_decoder_layers": 12,23 "num_heads": 12,24 "num_layers": 12,25 "output_past": true,26 "pad_token_id": 0,27 "quantization_config": {28 "_load_in_4bit": true,29 "_load_in_8bit": false,30 "bnb_4bit_compute_dtype": "float16",31 "bnb_4bit_quant_storage": "uint8",32 "bnb_4bit_quant_type": "nf4",33 "bnb_4bit_use_double_quant": false,34 "llm_int8_enable_fp32_cpu_offload": false,35 "llm_int8_has_fp16_weight": false,36 "llm_int8_skip_modules": null,37 "llm_int8_threshold": 6.0,38 "load_in_4bit": true,39 "load_in_8bit": false,40 "quant_method": "bitsandbytes"41 },42 "relative_attention_max_distance": 128,43 "relative_attention_num_buckets": 32,44 "torch_dtype": "float32",45 "transformers_version": "4.39.2",46 "use_cache": true,47 "vocab_size": 3210048}49 