| { | |
| "d_ff": 4096, | |
| "d_kv": 64, | |
| "d_model": 1024, | |
| "dropout_rate": 0.1, | |
| "finetuning_task": null, | |
| "initializer_factor": 1.0, | |
| "is_decoder": false, | |
| "layer_norm_epsilon": 1e-06, | |
| "n_positions": 512, | |
| "num_heads": 16, | |
| "num_labels": 2, | |
| "num_layers": 24, | |
| "output_attentions": false, | |
| "output_hidden_states": false, | |
| "output_past": true, | |
| "pruned_heads": {}, | |
| "relative_attention_num_buckets": 32, | |
| "torchscript": false, | |
| "use_bfloat16": false, | |
| "vocab_size": 32128 | |
| } | |