| { | |
| "dim": 1280, | |
| "n_layer": 30, | |
| "n_head": 20, | |
| "n_kv_head": null, | |
| "multiple_of": 256, | |
| "ffn_dim_multiplier": null, | |
| "rope_base": 10000, | |
| "norm_eps": 1e-05, | |
| "initializer_range": 0.02, | |
| "token_dropout_p": 0.1, | |
| "attn_dropout_p": 0.0, | |
| "resid_dropout_p": 0.1, | |
| "ffn_dropout_p": 0.1, | |
| "drop_path_rate": 0.0, | |
| "num_classes": 101, | |
| "class_dropout_prob": 0.1, | |
| "model_type": "class_cond", | |
| "vocab_size": 8192, | |
| "cls_token_num": 1, | |
| "max_batch_size": 32, | |
| "max_seq_len": 2048, | |
| "use_fixed_pe": false, | |
| "frame_prediction": false, | |
| "mode": "train_adap_ar", | |
| "with_block_sep": true, | |
| "tot_blocks": 4, | |
| "custom_cls_token_num": false, | |
| "start_ar_block_idx": 0 | |
| } |