| { |
| "architectures": [ |
| "YENForCausalLM" |
| ], |
| "model_type": "yen", |
| "torch_dtype": "float32", |
| "transformers_version": "4.49.0.dev0", |
| "vocab_size": 256, |
| "hidden_size": 1, |
| "num_hidden_layers": 1, |
| "num_attention_heads": 1, |
| "layer_norm_eps": 1e-06, |
| "max_position_embeddings": 15000000, |
| "rope_theta": 10000000.0, |
| "rope_scaling": { |
| "type": "dynamic", |
| "factor": 2.0 |
| }, |
| "activation_function": "gelu", |
| "tie_word_embeddings": false, |
| "use_cache": true, |
| "intermediate_size": 512, |
| "classifier_dropout": 0.1, |
| "attention_dropout": 0.1, |
| "bos_token_id": 0, |
| "eos_token_id": 255 |
| } |
