hyper-accel/tiny-random-starcoder2

1{2  "_attn_implementation_autoset": false,3  "_name_or_path": "bigcode/starcoder2-15b",4  "architectures": [5    "Starcoder2ForCausalLM"6  ],7  "attention_dropout": 0.1,8  "bos_token_id": 0,9  "embedding_dropout": 0.1,10  "eos_token_id": 0,11  "hidden_act": "gelu_pytorch_tanh",12  "hidden_size": 1024,13  "initializer_range": 0.01275,14  "intermediate_size": 4096,15  "max_position_embeddings": 16384,16  "mlp_type": "default",17  "model_type": "starcoder2",18  "norm_epsilon": 1e-05,19  "norm_type": "layer_norm",20  "num_attention_heads": 8,21  "num_hidden_layers": 2,22  "num_key_value_heads": 2,23  "residual_dropout": 0.1,24  "rope_scaling": null,25  "rope_theta": 100000,26  "sliding_window": 4096,27  "tie_word_embeddings": false,28  "torch_dtype": "float32",29  "transformers_version": "4.44.0",30  "use_bias": true,31  "use_cache": true,32  "vocab_size": 4915233}34