| { | |
| "architectures": [ | |
| "YsnrfdForCausalLM" | |
| ], | |
| "attention_bias": false, | |
| "attention_dropout": 0.0, | |
| "bos_token_id": 50256, | |
| "dtype": "float32", | |
| "early_stopping_patience": 3, | |
| "eos_token_id": 50256, | |
| "gradient_accumulation_steps": 4, | |
| "gradient_checkpointing": false, | |
| "hidden_dropout_prob": 0.0, | |
| "hidden_size": 768, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 3072, | |
| "max_position_embeddings": 2048, | |
| "model_type": "ysnrfd", | |
| "num_attention_heads": 12, | |
| "num_hidden_layers": 12, | |
| "pad_token_id": 50256, | |
| "rms_norm_eps": 1e-06, | |
| "rope_theta": 10000.0, | |
| "transformers_version": "4.57.1", | |
| "use_cache": true, | |
| "use_flash_attn": false, | |
| "use_gradient_accumulation": true, | |
| "vocab_size": 50257 | |
| } | |