| { |
| "_name_or_path": "openbmb/ProSparse-MiniCPM-1B-sft-llama-format", |
| "architectures": [ |
| "LlamaForCausalLM" |
| ], |
| "auto_map": { |
| "AutoConfig": "LlamaConfig", |
| "AutoModel": "LlamaModel", |
| "AutoModelForCausalLM": "LlamaForCausalLM", |
| "AutoModelForSeq2SeqLM": "LlamaForCausalLM", |
| "AutoModelForSequenceClassification": "LlamaForSequenceClassification" |
| }, |
| "bos_token_id": 1, |
| "eos_token_id": 2, |
| "hidden_act": "relu", |
| "hidden_size": 1536, |
| "initializer_range": 0.1, |
| "intermediate_size": 3840, |
| "max_position_embeddings": 4096, |
| "model_type": "llama", |
| "num_attention_heads": 24, |
| "num_hidden_layers": 52, |
| "num_key_value_heads": 8, |
| "rms_norm_eps": 1e-05, |
| "rope_scaling": null, |
| "torch_dtype": "bfloat16", |
| "transformers_version": "4.36.0", |
| "use_cache": true, |
| "vocab_size": 73440, |
| "scale_emb": 12, |
| "dim_model_base": 256, |
| "scale_depth": 1.4, |
| "rope_theta" : 10000 |
| } |