Text2Text Generation
Transformers
PyTorch
English
switch_transformers
switch-xxl-128 / config.json
ybelkada's picture
config.json (#2)
13cda2d
raw
history blame
1.84 kB
{
"architectures": [
"SwitchTransformersForConditionalGeneration"
],
"batch_prioritized_routing": false,
"d_ff": 10240,
"d_kv": 64,
"d_model": 4096,
"decoder_sparse_step": 2,
"dense_act_fn": "gelu",
"dropout_rate": 0.1,
"encoder_sparse_step": 2,
"eos_token_id": 1,
"expert_capacity": 64,
"feed_forward_proj": "gelu",
"initializer_factor": 1.0,
"decoder_start_token_id": 0,
"is_encoder_decoder": true,
"is_gated_act": false,
"layer_norm_epsilon": 1e-06,
"model_type": "switch_transformers",
"num_decoder_layers": 12,
"num_experts": 128,
"num_heads": 64,
"num_layers": 48,
"num_selected_experts": 1,
"num_sparse_decoder_layers": 12,
"num_sparse_encoder_layers": 12,
"pad_token_id": 0,
"relative_attention_max_distance": 128,
"relative_attention_num_buckets": 32,
"router_bias": false,
"router_dtype": "float32",
"router_ignore_padding_tokens": false,
"router_jitter_noise": 0,
"router_type": "tokens_masked",
"task_specific_params": {
"summarization": {
"early_stopping": true,
"length_penalty": 2.0,
"max_length": 200,
"min_length": 30,
"no_repeat_ngram_size": 3,
"num_beams": 4,
"prefix": "summarize: "
},
"translation_en_to_de": {
"early_stopping": true,
"max_length": 300,
"num_beams": 4,
"prefix": "translate English to German: "
},
"translation_en_to_fr": {
"early_stopping": true,
"max_length": 300,
"num_beams": 4,
"prefix": "translate English to French: "
},
"translation_en_to_ro": {
"early_stopping": true,
"max_length": 300,
"num_beams": 4,
"prefix": "translate English to Romanian: "
}
},
"torch_dtype": "bfloat16",
"transformers_version": "4.24.0.dev0",
"use_cache": true,
"vocab_size": 32128
}