Falconsai/topic_change_point
322
1{2 "_name_or_path": "./hf_model/",3 "architectures": [4 "T5ForConditionalGeneration"5 ],6 "classifier_dropout": 0.0,7 "d_ff": 2048,8 "d_kv": 64,9 "d_model": 512,10 "decoder_start_token_id": 0,11 "dense_act_fn": "relu",12 "dropout_rate": 0.1,13 "early_stopping": true,14 "eos_token_id": 1,15 "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 "length_penalty": 2.0,21 "max_length": 200,22 "min_length": 30,23 "model_type": "t5",24 "n_positions": 512,25 "no_repeat_ngram_size": 3,26 "num_beams": 4,27 "num_decoder_layers": 6,28 "num_heads": 8,29 "num_layers": 6,30 "output_past": true,31 "pad_token_id": 0,32 "prefix": "summarize: ",33 "relative_attention_max_distance": 128,34 "relative_attention_num_buckets": 32,35 "task_specific_params": {36 "summarization": {37 "early_stopping": true,38 "length_penalty": 2.0,39 "max_length": 200,40 "min_length": 30,41 "no_repeat_ngram_size": 3,42 "num_beams": 4,43 "prefix": "summarize: "44 },45 "translation_en_to_de": {46 "early_stopping": true,47 "max_length": 300,48 "num_beams": 4,49 "prefix": "translate English to German: "50 },51 "translation_en_to_fr": {52 "early_stopping": true,53 "max_length": 300,54 "num_beams": 4,55 "prefix": "translate English to French: "56 },57 "translation_en_to_ro": {58 "early_stopping": true,59 "max_length": 300,60 "num_beams": 4,61 "prefix": "translate English to Romanian: "62 }63 },64 "torch_dtype": "float32",65 "transformers_version": "4.41.2",66 "use_cache": true,67 "vocab_size": 3212868}69 