hf-tiny-model-private/tiny-random-SwitchTransformersForConditionalGeneration
058
1{2 "add_router_probs": false,3 "architectures": [4 "SwitchTransformersForConditionalGeneration"5 ],6 "bos_token_id": 0,7 "d_ff": 37,8 "d_kv": 8,9 "d_model": 32,10 "decoder_sparse_step": 1,11 "decoder_start_token_id": 0,12 "dense_act_fn": "relu",13 "dropout_rate": 0.1,14 "encoder_sparse_step": 1,15 "eos_token_id": 1,16 "expert_capacity": 100,17 "feed_forward_proj": "relu",18 "initializer_factor": 0.002,19 "is_encoder_decoder": true,20 "is_gated_act": false,21 "layer_norm_epsilon": 1e-06,22 "model_type": "switch_transformers",23 "num_decoder_layers": 5,24 "num_experts": 8,25 "num_heads": 4,26 "num_layers": 5,27 "num_sparse_decoder_layers": 3,28 "num_sparse_encoder_layers": 3,29 "pad_token_id": 0,30 "relative_attention_max_distance": 128,31 "relative_attention_num_buckets": 8,32 "router_aux_loss_coef": 0.001,33 "router_bias": false,34 "router_dtype": "float32",35 "router_ignore_padding_tokens": false,36 "router_jitter_noise": 0.0,37 "router_z_loss_coef": 0.001,38 "torch_dtype": "float32",39 "transformers_version": "4.28.0.dev0",40 "use_cache": true,41 "vocab_size": 3210042}43 