metadata
language: en
tags:
- clinical-nlp
- text-similarity
- regression
- bert
base_model: cross-encoder/monoelectra-large
Clinical STS Regression Model
Fine-tuned from cross-encoder/monoelectra-large on a clinical text-pair similarity dataset.
Training config
{
"train": "./med_sts/clinicalSTS.train.txt",
"test": "./med_sts/clinicalSTS.test.txt",
"model": [
"cross-encoder/monoelectra-large"
],
"epochs": 20,
"batch_size": 16,
"lr": 2e-05,
"weight_decay": 0.01,
"warmup_ratio": 0.1,
"max_grad_norm": 1.0,
"dropout": 0.1,
"max_len": 256,
"seed": 42,
"patience": 5,
"output_dir": "./sts_val/monoelectra-large",
"wandb_project": "clinical-sts",
"hf_repo": "youssefkhalil320/monoelectra-large"
}