| { | |
| "vocab_size": 32768, | |
| "dim": 800, | |
| "depth": 16, | |
| "n_heads": 10, | |
| "mlp_hidden": 2176, | |
| "depth_rank": 200, | |
| "depth_heads": 5, | |
| "seq_len": 4096, | |
| "max_context": 8192, | |
| "group_chunk": 128, | |
| "group_keep": 4, | |
| "initial_mix": 0.18, | |
| "recency_bias": 0.3, | |
| "rope_base": 500000.0, | |
| "norm_eps": 1e-05, | |
| "architectures": [ | |
| "CubicHierLM" | |
| ], | |
| "model_type": "cubic_hier_v7_gemma4", | |
| "torch_dtype": "bfloat16", | |
| "tie_word_embeddings": true, | |
| "parameter_count": 157065144 | |
| } |