| { |
| "activation_function": "relu", |
| "add_cross_attention": false, |
| "architectures": [ |
| "GPTBigCodeForCausalLM" |
| ], |
| "attention_softmax_in_fp32": false, |
| "attn_pdrop": 0.1, |
| "bos_token_id": 49152, |
| "dtype": "float32", |
| "embd_pdrop": 0.1, |
| "eos_token_id": 49152, |
| "gradient_checkpointing": false, |
| "initializer_range": 0.02, |
| "is_decoder": true, |
| "layer_norm_epsilon": 1e-05, |
| "model_type": "gpt_bigcode", |
| "multi_query": true, |
| "n_embd": 32, |
| "n_head": 4, |
| "n_inner": 37, |
| "n_layer": 2, |
| "n_positions": 512, |
| "num_key_value_heads": 1, |
| "pad_token_id": 49154, |
| "reorder_and_upcast_attn": false, |
| "resid_pdrop": 0.1, |
| "scale_attention_softmax_in_fp32": false, |
| "scale_attn_by_inverse_layer_idx": false, |
| "scale_attn_weights": true, |
| "tie_word_embeddings": true, |
| "transformers_version": "5.16.0.dev0", |
| "type_vocab_size": 16, |
| "use_cache": true, |
| "vocab_size": 49157 |
| } |
|
|
