reaperdoesntknow/DNA-50M

1{2  "alpha_init": 1.12,3  "architectures": [4    "LiquidForCausalLM"5  ],6  "attn_drop": 0.1,7  "bos_token_id": 50256,8  "chaos_scale": 0.2,9  "chead_kernal_size": 5,10  "cry_depth": 6,11  "cry_groups": 2,12  "cry_heads": 8,13  "cry_layers": 2,14  "dim": 256,15  "drop": 0.1,16  "dtype": "float32",17  "eos_token_id": 2,18  "feature_heads": 8,19  "freeze_thres": 0.5524292423753732,20  "gas_thres": 2.429242375373078e-06,21  "gnn_n_heads": 4,22  "gnn_num_layers": 4,23  "hidden": 256,24  "learn_alpha": true,25  "maha_init": 1.21,26  "max_position_embeddings": 512,27  "memory_size": 256,28  "metric": "maha_diag",29  "model_type": "liquid-former",30  "n_attn_heads": 16,31  "n_delta_windows": 5,32  "n_thought_heads": 32,33  "n_wind_size": 5,34  "num_forms": 2,35  "num_heads": 16,36  "num_hidden_layers": 1,37  "num_layers": 1,38  "pad_token_id": 0,39  "proj_drop": 0.1,40  "router_gate_heads": 16,41  "seq_init": 0.0002,42  "seq_len": 512,43  "share_kv": true,44  "t_step": 2,45  "temperature": null,46  "thermo_amp": 0.5,47  "thermo_freq": 1.1,48  "thought_lr": 0.0005,49  "thought_temp": 0.898,50  "tie_weights": true,51  "transformers_version": "4.56.1",52  "vocab_size": 5025753}54