ybelkada/tiny-random-T5ForConditionalGeneration-calibrated
1178k
1{2 "_name_or_path": "hf-internal-testing/tiny-random-T5ForConditionalGeneration",3 "architectures": [4 "T5ForConditionalGeneration"5 ],6 "bos_token_id": 0,7 "d_ff": 37,8 "d_kv": 8,9 "d_model": 32,10 "decoder_start_token_id": 0,11 "dense_act_fn": "relu",12 "dropout_rate": 0.1,13 "eos_token_id": 1,14 "feed_forward_proj": "relu",15 "initializer_factor": 0.2,16 "is_encoder_decoder": true,17 "is_gated_act": false,18 "layer_norm_epsilon": 1e-06,19 "model_type": "t5",20 "num_decoder_layers": 5,21 "num_heads": 4,22 "num_layers": 5,23 "pad_token_id": 0,24 "relative_attention_max_distance": 128,25 "relative_attention_num_buckets": 8,26 "torch_dtype": "float32",27 "transformers_version": "4.28.0.dev0",28 "use_cache": true,29 "vocab_size": 3210030}31 