| { | |
| "model_type": "latex_decoder", | |
| "architectures": [ | |
| "LaTeXDecoderForCausalLM" | |
| ], | |
| "auto_map": { | |
| "AutoConfig": "configuration_latex_decoder.LaTeXDecoderConfig", | |
| "AutoModelForCausalLM": "modeling_latex_decoder.LaTeXDecoderForCausalLM" | |
| }, | |
| "vocab_size": 2046, | |
| "pad_id": 0, | |
| "bos_id": 2, | |
| "eos_id": 3, | |
| "pad_token_id": 0, | |
| "bos_token_id": 2, | |
| "eos_token_id": 3, | |
| "d_model": 512, | |
| "n_heads": 8, | |
| "n_layers": 6, | |
| "d_ff": 1408, | |
| "dropout": 0.1, | |
| "max_seq_len": 200, | |
| "rope_theta": 10000.0, | |
| "tie_weights": true, | |
| "pretrain_step": 56000, | |
| "pretrain_loss": 1.100601 | |
| } |