{ "_name_or_path": "/mnt/hdfs/caozy/SFT_model", "architectures": [ "Olmo2ForCausalLMForSemiNAT" ], "attention_bias": false, "attention_dropout": 0.0, "chunk_size_limit": 5, "decoder_layers": 9, "encoder_layer": 1, "eos_token_id": 100257, "hidden_act": "silu", "hidden_size": 2048, "initializer_range": 0.02, "intermediate_size": 8192, "layers": 9, "length_loss_type": "ce", "max_position_embeddings": 4096, "mlp": false, "model_type": "olmo2", "num_attention_heads": 16, "num_hidden_layers": 9, "num_key_value_heads": 16, "pad_token_id": 100277, "position_embedding_type": "absolute", "rms_norm_eps": 1e-06, "rope_scaling": null, "rope_theta": 500000, "tie_word_embeddings": false, "torch_dtype": "bfloat16", "transformers_version": "4.49.0", "use_cache": true, "vocab_size": 100352 }