hf-tiny-v2/tiny-random-GPTBigCodeForCausalLM

1{2  "activation_function": "relu",3  "add_cross_attention": false,4  "architectures": [5    "GPTBigCodeForCausalLM"6  ],7  "attention_softmax_in_fp32": false,8  "attn_pdrop": 0.1,9  "bos_token_id": 49152,10  "dtype": "float32",11  "embd_pdrop": 0.1,12  "eos_token_id": 49152,13  "gradient_checkpointing": false,14  "initializer_range": 0.02,15  "is_decoder": true,16  "layer_norm_epsilon": 1e-05,17  "model_type": "gpt_bigcode",18  "multi_query": true,19  "n_embd": 32,20  "n_head": 4,21  "n_inner": 37,22  "n_layer": 2,23  "n_positions": 512,24  "num_key_value_heads": 1,25  "pad_token_id": 49154,26  "reorder_and_upcast_attn": false,27  "resid_pdrop": 0.1,28  "scale_attention_softmax_in_fp32": false,29  "scale_attn_by_inverse_layer_idx": false,30  "scale_attn_weights": true,31  "tie_word_embeddings": true,32  "transformers_version": "5.16.0.dev0",33  "type_vocab_size": 16,34  "use_cache": true,35  "vocab_size": 4915736}37