| { | |
| "reader_layers": [ | |
| 24, | |
| 25, | |
| 26, | |
| 27, | |
| 28, | |
| 29, | |
| 30, | |
| 31 | |
| ], | |
| "write_layer": 4, | |
| "cfg": { | |
| "vocab_size": 100278, | |
| "d_model": 4096, | |
| "n_layers": 32, | |
| "n_heads": 32, | |
| "n_experts": 1, | |
| "top_k": 1, | |
| "ff_mult": 2, | |
| "ctx_len": 4096, | |
| "graph_slots": 8, | |
| "read_heads": 1, | |
| "write_layer": 0, | |
| "write_span_len": 16, | |
| "lb_coef": 0.01, | |
| "read_chunk": 256, | |
| "tie_embeddings": false, | |
| "reader_init_std": 0.02 | |
| } | |
| } |