| { |
| "model_type": "cosmicfish_hrm", |
| "architectures": [ |
| "HRMCosmicFish" |
| ], |
| "vocab_size": 50304, |
| "n_embd": 448, |
| "block_size": 512, |
| "n_head": 8, |
| "n_kv_head": 4, |
| "n_input_layers": 6, |
| "n_output_layers": 6, |
| "hrm_H_layers": 4, |
| "hrm_L_layers": 4, |
| "hrm_H_cycles": 2, |
| "hrm_L_cycles": 2, |
| "hrm_max_steps": 16, |
| "hrm_exploration_prob": 0.05, |
| "dropout": 0.1, |
| "bias": false, |
| "use_rotary": true, |
| "use_gqa": true, |
| "use_swiglu": true, |
| "eps": 1e-05, |
| "torch_dtype": "float32", |
| "transformers_version": "4.41.0", |
| "pad_token_id": 50256, |
| "bos_token_id": 50256, |
| "eos_token_id": 50256 |
| } |