| { | |
| "P_rank": 1, | |
| "architectures": [ | |
| "HamiltonForCausalLM" | |
| ], | |
| "attn": null, | |
| "attn_mode": "chunk", | |
| "bos_token_id": 1, | |
| "conv_bias_spatial": true, | |
| "conv_bias_temporal": true, | |
| "conv_size_spatial": 4, | |
| "conv_size_temporal": 16, | |
| "decay_weight": 4.0, | |
| "dropout": 0.0, | |
| "dtype": "float32", | |
| "eos_token_id": 2, | |
| "expand_spatial": 4, | |
| "fuse_cross_entropy": true, | |
| "fuse_norm": true, | |
| "fuse_swiglu": true, | |
| "head_dim": 64, | |
| "hidden_act": "swish", | |
| "hidden_ratio": 3, | |
| "hidden_size": 1024, | |
| "initializer_range": 0.02, | |
| "intermediate_size": null, | |
| "max_position_embeddings": 8192, | |
| "mnu_bias": true, | |
| "model_type": "hamilton", | |
| "norm_eps": 1e-06, | |
| "num_heads": 16, | |
| "num_hidden_layers": 24, | |
| "task": "text", | |
| "tie_word_embeddings": false, | |
| "transformers_version": "4.57.3", | |
| "use_cache": true, | |
| "use_l2warp": false, | |
| "use_mlp": true, | |
| "vocab_size": 32000 | |
| } | |