| { | |
| "_name_or_path": "/data1/hrsun/FuxiTranyu-8B", | |
| "activation_function": "gelu_fast", | |
| "architectures": [ | |
| "FuxiTranyuForCausalLM" | |
| ], | |
| "attention_bias": true, | |
| "attention_dropout": 0.0, | |
| "auto_map": { | |
| "AutoConfig": "configuration_fuxitranyu.FuxiTranyuConfig", | |
| "AutoModelForCausalLM": "modeling_fuxitranyu.FuxiTranyuForCausalLM" | |
| }, | |
| "bos_token_id": 0, | |
| "eos_token_id": 0, | |
| "hidden_size": 4096, | |
| "initializer_range": 0.01, | |
| "intermediate_size": 16384, | |
| "max_position_embeddings": 4096, | |
| "model_type": "fuxitranyu", | |
| "num_attention_heads": 32, | |
| "num_hidden_layers": 30, | |
| "num_key_value_heads": 32, | |
| "pretraining_tp": 1, | |
| "rms_norm_eps": 1e-05, | |
| "rope_scaling": null, | |
| "rope_theta": 10000.0, | |
| "tie_word_embeddings": false, | |
| "tokenizer_class": "GPT2TokenizerFast", | |
| "torch_dtype": "float16", | |
| "transformers_version": "4.42.4", | |
| "use_cache": true, | |
| "vocab_size": 250680 | |
| } | |