| { | |
| "bias": false, | |
| "block_size": 1023, | |
| "dropout": 0.0, | |
| "expert_head": 8, | |
| "n_embd": 512, | |
| "n_experts": 16384, | |
| "n_experts_per_head": 16, | |
| "n_head": 8, | |
| "n_layer": 8, | |
| "vocab_size": 32 | |
| } |
| { | |
| "bias": false, | |
| "block_size": 1023, | |
| "dropout": 0.0, | |
| "expert_head": 8, | |
| "n_embd": 512, | |
| "n_experts": 16384, | |
| "n_experts_per_head": 16, | |
| "n_head": 8, | |
| "n_layer": 8, | |
| "vocab_size": 32 | |
| } |