| { | |
| "architectures": [ | |
| "VerMindDenseForCausalLM" | |
| ], | |
| "attention_bias": false, | |
| "attention_dropout": 0.0, | |
| "dtype": "float16", | |
| "eos_token_id": 32769, | |
| "head_dim": 64, | |
| "hidden_dropout": 0.0, | |
| "hidden_size": 768, | |
| "intermediate_size": 2064, | |
| "max_position_embeddings": 2048, | |
| "model_type": "vermind_dense", | |
| "num_attention_heads": 12, | |
| "num_hidden_layers": 12, | |
| "num_key_value_heads": 4, | |
| "pad_token_id": 0, | |
| "residual_dropout": 0.0, | |
| "rms_norm_eps": 1e-06, | |
| "rope_scaling": { | |
| "attention_factor": 1.0, | |
| "beta_fast": 32.0, | |
| "beta_slow": 1.0, | |
| "factor": 4.0, | |
| "original_max_position_embeddings": 2048, | |
| "type": "yarn" | |
| }, | |
| "rope_theta": 10000.0, | |
| "transformers_version": "4.57.5", | |
| "use_cache": false, | |
| "use_flash_attention": false, | |
| "vocab_size": 32772 | |
| } | |