minijev-general / config.json
markuz89's picture
MiniJEV v2 model export
45507bb verified
Raw History Blame Contribute Delete
1.62 kB
{
"format": "minijev-v2",
"config": {
"seed": 0,
"task": {
"name": "text",
"max_len": 32
},
"data": {
"dir": "data/general_qa_teacher"
},
"model": {
"d_model": 128,
"n_heads": 4,
"n_layers": 2,
"n_decision_layers": 1,
"mlp_ratio": 2.0,
"dropout": 0.1,
"max_question_len": 48,
"max_choice_len": 16,
"scorer": "bi",
"text_encoder": {
"type": "pretrained",
"name": "text_encoder",
"finetune": "full",
"tokenizer": "text_encoder"
},
"state_encoders": {
"text": {}
}
},
"training": {
"batch_size": 32,
"epochs": 6,
"lr": 0.001,
"min_lr_ratio": 0.05,
"weight_decay": 0.05,
"warmup_steps": 100,
"grad_clip": 1.0,
"choice_dropout": 0.0,
"eval_every": 500,
"checkpoint_every": 1000,
"log_every": 100,
"cache_text": true,
"hardware": {
"device": "auto",
"amp": "auto",
"num_workers": 0,
"compile": false
},
"backbone_lr": 3e-05
},
"loss": {
"hard_weight": 1.0,
"distillation": {
"enabled": true,
"alpha": 0.5,
"temperature": 1.0
}
},
"calibration": {
"method": "temperature"
},
"name": "qat-minilm-tiny-bi-full"
},
"temperature": 3.506696939468384,
"task_temperatures": {
"text": 3.506696820258835
},
"metrics": {
"test_acc_teacher": 0.812,
"test_ece_teacher": 0.055,
"test_acc_knowledge_table": 0.827,
"benchmark_30": "23/30"
}
}