| { | |
| "alpha_init": 1.5, | |
| "architectures": [ | |
| "LiquidForCausalLM" | |
| ], | |
| "attn_drop": 0.1, | |
| "bos_token_id": null, | |
| "chead_kernal_size": 5, | |
| "dim": 256, | |
| "drop": 0.1, | |
| "dtype": "float32", | |
| "eos_token_id": 2, | |
| "feature_heads": 64, | |
| "gnn_n_heads": 4, | |
| "gnn_num_layers": 1, | |
| "hidden": 256, | |
| "learn_alpha": true, | |
| "maha_init": 1.6, | |
| "max_position_embeddings": 512, | |
| "memory_size": 256, | |
| "metric": "maha_diag", | |
| "model_type": "liquid-former", | |
| "n_attn_heads": 8, | |
| "n_delta_windows": 10, | |
| "n_thought_heads": 64, | |
| "num_heads": 32, | |
| "num_layers": 1, | |
| "pad_token_id": 0, | |
| "proj_drop": 0.1, | |
| "router_gate_heads": 32, | |
| "share_kv": true, | |
| "tie_weights": true, | |
| "transformers_version": "4.56.1", | |
| "vocab_size": 151669 | |
| } | |