| { | |
| "architectures": ["TransfoXLModel"], | |
| "model_type": "transfo-xl", | |
| "n_token": 267735, | |
| "d_embed": 512, | |
| "d_model": 512, | |
| "d_head": 64, | |
| "d_inner": 2048, | |
| "n_head": 8, | |
| "n_layer": 6, | |
| "mem_len": 512, | |
| "same_length": false, | |
| "clamp_len": -1, | |
| "dropout": 0.1, | |
| "dropatt": 0.1 | |
| } |