amd/Instella-3B-Instruct

1{2  "architectures": [3    "InstellaForCausalLM"4  ],5  "attention_bias": false,6  "attention_dropout": 0.0,7  "auto_map": {8    "AutoConfig": "modeling_instella.InstellaConfig",9    "AutoModelForCausalLM": "modeling_instella.InstellaForCausalLM"10  },11  "bos_token_id": 0,12  "eos_token_id": 0,13  "hidden_act": "silu",14  "hidden_size": 2560,15  "initializer_range": 0.02,16  "intermediate_size": 6912,17  "max_position_embeddings": 4096,18  "model_type": "instella",19  "num_attention_heads": 32,20  "num_hidden_layers": 36,21  "num_key_value_heads": 32,22  "pad_token_id": 1,23  "rms_norm_eps": 1e-05,24  "rope_scaling": null,25  "rope_theta": 10000.0,26  "tie_word_embeddings": false,27  "torch_dtype": "bfloat16",28  "transformers_version": "4.48.0",29  "use_cache": true,30  "vocab_size": 5030431}32