| { | |
| "d_conv": 4, | |
| "d_model": 512, | |
| "d_state": 16, | |
| "dropout": 0.1, | |
| "expand": 2, | |
| "gradient_checkpointing": true, | |
| "max_position_embeddings": 512, | |
| "n_layer": 12, | |
| "pad_token_id": 0, | |
| "vocab_size": 64000 | |
| } |
| { | |
| "d_conv": 4, | |
| "d_model": 512, | |
| "d_state": 16, | |
| "dropout": 0.1, | |
| "expand": 2, | |
| "gradient_checkpointing": true, | |
| "max_position_embeddings": 512, | |
| "n_layer": 12, | |
| "pad_token_id": 0, | |
| "vocab_size": 64000 | |
| } |