ca-byte-mt-en2x / config.json
sujayrittikar's picture
CA byte-LM translator (en2x)
1013554 verified
Raw History Blame Contribute Delete
1.05 kB
{
"architecture": "CAByteLM",
"task": "translation",
"direction": "en2x",
"vocab": 266,
"lang_tag": {
"English": 259,
"Assamese": 260,
"Khasi": 261,
"Manipuri": 262,
"Meitei-Mayek": 263,
"Mizo": 264,
"Nyishi": 265
},
"FRONTEND": "nca",
"SMOKE_TEST": false,
"seed": 42,
"langs": [
"Assamese",
"Khasi",
"Manipuri",
"Meitei-Mayek",
"Mizo",
"Nyishi"
],
"pretrain_english": true,
"data_dir": "data_clean",
"mono_dir": "mono_data",
"use_mono": true,
"out_dir": "runs/ca_ncabig",
"d_model": 768,
"n_layers": 8,
"n_heads": 12,
"d_ff": 3072,
"dropout": 0.1,
"seq_len": 512,
"nca_steps": 12,
"nca_hidden": 1536,
"nca_kernel": 5,
"nca_fire_rate": 0.5,
"batch_size": 32,
"grad_accum": 2,
"lr": 0.0003,
"min_lr_ratio": 0.1,
"warmup_steps": 500,
"weight_decay": 0.1,
"betas": [
0.9,
0.95
],
"grad_clip": 1.0,
"max_steps": 40000,
"gpu": 0,
"eval_every": 1000,
"gen_every": 2000,
"n_val_blocks": 40,
"ckpt_every": 2000
}