| { | |
| "_description": "Deep-Narrow (Paper Deep): 30 layers \u00d7 384 hidden \u00d7 1024 intermediate - ~59.50M parameters. Paper SOTA configuration from original paper's 125M deep architecture.", | |
| "_name_or_path": "/home/yaliangcai/mobilelm/MobileLLM/configs/60M_deep_scale/", | |
| "architectures": [ | |
| "LlamaForCausalLM" | |
| ], | |
| "attention_bias": false, | |
| "attention_dropout": 0.0, | |
| "bos_token_id": 1, | |
| "eos_token_id": 2, | |
| "hidden_act": "silu", | |
| "hidden_size": 384, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 1024, | |
| "layer_sharing": true, | |
| "max_position_embeddings": 2048, | |
| "mlp_bias": false, | |
| "model_type": "llama", | |
| "num_attention_heads": 6, | |
| "num_hidden_layers": 30, | |
| "num_key_value_heads": 2, | |
| "pretraining_tp": 1, | |
| "rms_norm_eps": 1e-05, | |
| "rope_scaling": null, | |
| "rope_theta": 10000.0, | |
| "share_embedding": true, | |
| "tie_word_embeddings": false, | |
| "torch_dtype": "float32", | |
| "transformers_version": "4.41.2", | |
| "use_cache": true, | |
| "vocab_size": 32000 | |
| } | |