| { |
| "_name_or_path": "Xylaria", |
| "architectures": [ |
| "XylariaTransformer" |
| ], |
| "attention_dropout": 0.003, |
| "bos_token_id": 151643, |
| "custom_attention_type": "grouped_query", |
| "custom_settings": { |
| "param_0": 0.18909638773500503, |
| "param_1": 0.8193609427023019, |
| "param_2": 0.8222061374701647, |
| "param_3": 0.9849500721409111, |
| "param_4": 0.6006148886232605, |
| "param_5": 0.27476915865965346, |
| "param_6": 0.38785891496724245, |
| "param_7": 0.4125778939436352, |
| "param_8": 0.5190071705230352, |
| "param_9": 0.8516028213154339 |
| }, |
| "embedding_dropout": 0.0165, |
| "eos_token_id": 151645, |
| "extra_layer_norm": false, |
| "hidden_act": "silu", |
| "hidden_size": 5119, |
| "initializer_range": 0.0208, |
| "intermediate_size": 27647, |
| "max_position_embeddings": 40960, |
| "max_window_layers": 64, |
| "model_type": "qwen2", |
| "num_attention_heads": 39, |
| "num_hidden_layers": 63, |
| "num_key_value_heads": 8, |
| "obfuscation_level": 3, |
| "rms_norm_eps": 1e-05, |
| "rope_scaling": null, |
| "rope_theta": 1028724.3021944767, |
| "seed": 1741770187, |
| "sliding_window": null, |
| "tie_word_embeddings": false, |
| "torch_dtype": "float32", |
| "transformers_version": "4.48.3", |
| "use_cache": true, |
| "use_sliding_window": false, |
| "vocab_size": 152064, |
| "xylaria_version": "1.8" |
| } |
|
|