{ "bos_token_id": 0, "d_ff": 37, "d_kv": 8, "d_model": 32, "decoder_start_token_id": 0, "dropout_rate": 0.1, "eos_token_id": 1, "feed_forward_proj": "relu", "gradient_checkpointing": false, "initializer_factor": 0.002, "is_encoder_decoder": true, "layer_norm_epsilon": 1e-06, "model_type": "t5", "num_decoder_layers": 5, "num_heads": 4, "num_layers": 5, "pad_token_id": 0, "relative_attention_num_buckets": 8, "transformers_version": "4.11.0.dev0", "use_cache": true, "vocab_size": 1103 }