aieng-lab/starcoder2-7b_incivility
05
1{2 "_name_or_path": "bigcode/starcoder2-7b",3 "activation_function": "gelu",4 "architectures": [5 "Starcoder2ForSequenceClassification"6 ],7 "attention_dropout": 0.1,8 "attention_softmax_in_fp32": true,9 "bos_token_id": 0,10 "embedding_dropout": 0.1,11 "eos_token_id": 0,12 "hidden_act": "gelu_pytorch_tanh",13 "hidden_size": 4608,14 "initializer_range": 0.018042,15 "intermediate_size": 18432,16 "layer_norm_epsilon": 1e-05,17 "max_position_embeddings": 16384,18 "mlp_type": "default",19 "model_type": "starcoder2",20 "norm_epsilon": 1e-05,21 "norm_type": "layer_norm",22 "num_attention_heads": 36,23 "num_hidden_layers": 32,24 "num_key_value_heads": 4,25 "pad_token_id": 0,26 "residual_dropout": 0.1,27 "rope_scaling": null,28 "rope_theta": 1000000,29 "scale_attention_softmax_in_fp32": true,30 "scale_attn_weights": true,31 "sliding_window": 4096,32 "torch_dtype": "bfloat16",33 "transformers_version": "4.48.3",34 "use_bias": true,35 "use_cache": true,36 "vocab_size": 4915237}38 