LevT-Nekoizer / config.json
KrisTHL181's picture
Upload LevT-Nekoizer model weights (step 300k, val_loss 5.087)
33ced16 verified
Raw
History Blame Contribute Delete
605 Bytes
{
"vocab_size": 73448,
"pad_token_id": 0,
"bos_token_id": 1,
"eos_token_id": 2,
"plh_token_id": 73446,
"embedding_dim": 1024,
"d_model": 512,
"n_heads": 8,
"d_ff": 2048,
"n_encoder_layers": 6,
"n_decoder_layers": 6,
"dropout": 0.0,
"attention_dropout": 0.0,
"activation": "gelu",
"pos_encoding_type": "alibi",
"rope_base": 10000.0,
"max_source_positions": 1024,
"max_target_positions": 1024,
"qk_norm": true,
"headwise_attn_output_gate": false,
"elementwise_attn_output_gate": false,
"max_placeholder": 255,
"early_exit_del": null,
"early_exit_plh": null
}