| { | |
| "_name_or_path": "/mnt/algorithm/user_dir/fangyue/.cache/huggingface/hub/models--OpenLLMAI--Llama-2-7b-sft-model-ocra-500k/snapshots/6fce8d06f40f2359ec447ab7f33333ffd2019b47", | |
| "architectures": [ | |
| "LlamaForCausalLM" | |
| ], | |
| "attention_bias": false, | |
| "attention_dropout": 0.0, | |
| "attention_probs_dropout_prob": 0, | |
| "bos_token_id": 1, | |
| "classifier_dropout_prob": 0.1, | |
| "embedding_size": 128, | |
| "eos_token_id": 2, | |
| "hidden_act": "silu", | |
| "hidden_dropout_prob": 0, | |
| "hidden_size": 4096, | |
| "initializer_range": 0.02, | |
| "inner_group_num": 1, | |
| "intermediate_size": 11008, | |
| "layer_norm_eps": 1e-12, | |
| "max_position_embeddings": 4096, | |
| "model_type": "albert", | |
| "num_attention_heads": 32, | |
| "num_hidden_groups": 1, | |
| "num_hidden_layers": 32, | |
| "num_key_value_heads": 32, | |
| "pad_token_id": 2, | |
| "position_embedding_type": "absolute", | |
| "pretraining_tp": 1, | |
| "rms_norm_eps": 1e-05, | |
| "rope_scaling": null, | |
| "rope_theta": 10000.0, | |
| "tie_word_embeddings": false, | |
| "torch_dtype": "bfloat16", | |
| "transformers_version": "4.40.0", | |
| "type_vocab_size": 2, | |
| "use_cache": true, | |
| "vocab_size": 32000 | |
| } | |