| { | |
| "vocab_size": 114, | |
| "hidden_size": 448, | |
| "num_hidden_layers": 16, | |
| "num_attention_heads": 7, | |
| "num_key_value_heads": 7, | |
| "intermediate_size": 1344, | |
| "max_position_embeddings": 8192, | |
| "rms_norm_eps": "1e-6", | |
| "initializer_range": 0.02, | |
| "use_cache": true, | |
| "num_experts": 8, | |
| "num_experts_per_tok": 2, | |
| "expert_capacity_factor": 1.5, | |
| "router_aux_loss_coef": 0.01, | |
| "moe_implementation": "megablocks", | |
| "moe_world_size": 4, | |
| "resid_dropout": 0.0, | |
| "hidden_dropout": 0.0, | |
| "dropout_warmup_steps": 0, | |
| "dropout_ramp_steps": 0, | |
| "dropout_schedule": "linear", | |
| "attention_dropout": 0.0, | |
| "gradient_clip_norm": 0.0, | |
| "label_smoothing": 0.0 | |
| } |