| { |
| "architectures": [ |
| "LongT5Model" |
| ], |
| "bos_token_id": 0, |
| "d_ff": 37, |
| "d_kv": 8, |
| "d_model": 32, |
| "decoder_start_token_id": 0, |
| "dense_act_fn": "relu", |
| "dropout_rate": 0.1, |
| "encoder_attention_type": "local", |
| "eos_token_id": 1, |
| "feed_forward_proj": "relu", |
| "global_block_size": 3, |
| "initializer_factor": 0.002, |
| "is_encoder_decoder": true, |
| "is_gated_act": false, |
| "layer_norm_epsilon": 1e-06, |
| "local_radius": 5, |
| "model_type": "longt5", |
| "num_decoder_layers": 5, |
| "num_heads": 4, |
| "num_layers": 5, |
| "pad_token_id": 0, |
| "relative_attention_max_distance": 128, |
| "relative_attention_num_buckets": 8, |
| "torch_dtype": "float32", |
| "transformers_version": "4.25.0.dev0", |
| "use_cache": true, |
| "vocab_size": 1302 |
| } |
|
|
