| { | |
| "dim": 1024, | |
| "n_layer": 24, | |
| "n_head": 16, | |
| "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": 1000, | |
| "caption_dim": 2048, | |
| "class_dropout_prob": 0.1, | |
| "model_type": "c2i", | |
| "vocab_size": 8192, | |
| "cls_token_num": 1, | |
| "block_size": 128, | |
| "max_batch_size": 32, | |
| "max_seq_len": 2048, | |
| "learnable_tok_embeddings": true, | |
| "group_size": 8, | |
| "causal_num": null | |
| } |