| { | |
| "auto_map": { | |
| "AutoConfig": "modeling_dense_gau_retnet.DenseGauRetNetConfig", | |
| "AutoModelForCausalLM": "modeling_dense_gau_retnet.DenseGauRetNetForCausalLM", | |
| "AutoModelForSequenceClassification": "modeling_dense_gau_retnet.DenseGauRetNetForSequenceClassification" | |
| }, | |
| "bos_token_id": 1, | |
| "eos_token_id": 2, | |
| "hidden_act": "silu", | |
| "hidden_size": 1536, | |
| "query_key_dim": 768, | |
| "initializer_range": 0.02, | |
| "max_position_embeddings": 2048, | |
| "model_type": "DenseGauRetNet", | |
| "num_attention_heads": 2, | |
| "num_hidden_layers": 16, | |
| "pad_token_id": 0, | |
| "rms_norm_eps": 1e-06, | |
| "layernorm_eps": 1e-5, | |
| "retnorm": false, | |
| "tie_word_embeddings": false, | |
| "torch_dtype": "float16", | |
| "transformers_version": "4.29.1", | |
| "use_cache": false, | |
| "vocab_size": 32001, | |
| "v_factor": 2, | |
| "intermediate_k_select_scale": 8, | |
| "intermediate_v_select_scale": 32, | |
| "dense_block_layers": 2, | |
| "dropout": 0.1, | |
| "deepnorm": false | |
| } |