FinText/Chronos_Small_2022_Global
0211
1{2 "_name_or_path": "google/t5-efficient-small",3 "architectures": [4 "T5ForConditionalGeneration"5 ],6 "chronos_config": {7 "context_length": 512,8 "eos_token_id": 1,9 "model_type": "seq2seq",10 "n_special_tokens": 2,11 "n_tokens": 4096,12 "num_samples": 20,13 "pad_token_id": 0,14 "prediction_length": 64,15 "temperature": 1.0,16 "tokenizer_class": "MeanScaleUniformBins",17 "tokenizer_kwargs": {18 "high_limit": 15.0,19 "low_limit": -15.020 },21 "top_k": 50,22 "top_p": 1.0,23 "use_eos_token": true24 },25 "classifier_dropout": 0.0,26 "d_ff": 2048,27 "d_kv": 64,28 "d_model": 512,29 "decoder_start_token_id": 0,30 "dense_act_fn": "relu",31 "dropout_rate": 0.1,32 "eos_token_id": 1,33 "feed_forward_proj": "relu",34 "initializer_factor": 0.05,35 "is_encoder_decoder": true,36 "is_gated_act": false,37 "layer_norm_epsilon": 1e-06,38 "model_type": "t5",39 "n_positions": 512,40 "num_decoder_layers": 6,41 "num_heads": 8,42 "num_layers": 6,43 "pad_token_id": 0,44 "relative_attention_max_distance": 128,45 "relative_attention_num_buckets": 32,46 "torch_dtype": "float32",47 "transformers_version": "4.41.0.dev0",48 "use_cache": true,49 "vocab_size": 409650}51 