| { | |
| "model_type": "tinyv4", | |
| "architectures": ["TinyV4"], | |
| "auto_map": { | |
| "AutoModel": "modeling_tinyv4.TinyV4", | |
| "AutoModelForCausalLM": "modeling_tinyv4.TinyV4" | |
| }, | |
| "vocab_size": 32000, | |
| "dim": 128, | |
| "depth": 6, | |
| "n_hc": 2, | |
| "n_routed": 4, | |
| "n_active": 2, | |
| "n_shared": 1, | |
| "expert_intermediate": 192, | |
| "csa_m": 4, | |
| "csa_topk": 16, | |
| "hca_m": 8, | |
| "n_win": 16, | |
| "n_q_head": 4, | |
| "head_dim": 48, | |
| "d_c": 64, | |
| "n_idx_head": 4, | |
| "idx_head_dim": 48, | |
| "n_out_group": 2, | |
| "d_g": 64, | |
| "rope_dim": 24, | |
| "mtp_depth": 1, | |
| "hash_layers": 2, | |
| "max_len": 512, | |
| "sinkhorn_iters": 20, | |
| "aux_bias_update": 0.001, | |
| "bal_loss_weight": 0.0001, | |
| "step": 5000, | |
| "best_loss": 3.9723572731018066 | |
| } | |