{ | |
"architectures": [ | |
"GPTSanJapaneseForConditionalGeneration" | |
], | |
"d_ext": 80, | |
"d_ff": 32, | |
"d_model": 32, | |
"d_spout": 33, | |
"dropout_rate": 0.0, | |
"eos_token_id": 35999, | |
"expert_capacity": 100, | |
"initializer_factor": 0.002, | |
"layer_norm_epsilon": 1e-06, | |
"max_position_embeddings": 1280, | |
"model_type": "gptsan-japanese", | |
"num_contexts": 7, | |
"num_experts": 2, | |
"num_ext_layers": 2, | |
"num_heads": 4, | |
"num_layers": 5, | |
"num_switch_layers": 3, | |
"output_router_logits": false, | |
"pad_token_id": 35999, | |
"router_bias": false, | |
"router_dtype": "float32", | |
"router_ignore_padding_tokens": false, | |
"router_jitter_noise": 0.0, | |
"separator_token_id": 35998, | |
"torch_dtype": "float32", | |
"transformers_version": "4.28.0.dev0", | |
"use_cache": true, | |
"vocab_size": 36000 | |
} | |