| { |
| "a_log_min": -3.912023005428146, |
| "architectures": [ |
| "EMG3ChForCausalLM" |
| ], |
| "auto_map": { |
| "AutoConfig": "emg_ssm_3channel.EMG3ChConfig", |
| "AutoModelForCausalLM": "emg_ssm_3channel.EMG3ChForCausalLM" |
| }, |
| "bos_token_id": 1, |
| "boundary_token_ids": [ |
| 47, |
| 57, |
| 60, |
| 70, |
| 81, |
| 88, |
| 23436, |
| 23438, |
| 23440, |
| 23442, |
| 23444, |
| 23446, |
| 23448, |
| 23450 |
| ], |
| "ce_chunk": 64, |
| "compose_on_close": true, |
| "cue_key_dim": 32, |
| "cue_val_dim": 64, |
| "d_channel": 128, |
| "d_state": 8, |
| "depth_target": 1.5, |
| "dropout": 0.1, |
| "dt_max": 2.0, |
| "dtype": "float32", |
| "eos_token_id": 2, |
| "force_stay": false, |
| "gumbel_temperature": 0.5, |
| "head_mode": "hier", |
| "head_shortlist": 4096, |
| "head_tail_bins": 3, |
| "hidden_size": 704, |
| "hier_K1": 4096, |
| "hier_K2": 128, |
| "hier_kernel": 3, |
| "hier_mid": 2048, |
| "hier_top": 128, |
| "k_params": 8, |
| "kill_A": false, |
| "kill_B": false, |
| "kill_D": false, |
| "lambda_balance": 0.05, |
| "lambda_boundary": 0.0, |
| "lambda_cpc": 0.1, |
| "lambda_depth": 0.4, |
| "lambda_load": 0.05, |
| "lambda_param_sup": 0.1, |
| "lambda_success": 0.0, |
| "load_target_open": 0.1, |
| "logit_cap": 6.0, |
| "mask_pad_loss": true, |
| "max_position_embeddings": 1024, |
| "model_type": "ssm_emg_3ch", |
| "n_phase": 6, |
| "n_slots": 4, |
| "pad_token_id": 3, |
| "param_film": true, |
| "param_label_smooth": 0.05, |
| "param_reg": 0.003, |
| "parametric_mode": "film", |
| "return_train_logits": false, |
| "seal_init": 0.85, |
| "slice_mode": "cue", |
| "stay_force": 0.0, |
| "tie_word_embeddings": true, |
| "transformers_version": "5.3.0", |
| "use_expectation": true, |
| "use_parameters": true, |
| "vocab_size": 39697 |
| } |
|
|