🚀 CS224n SQuAD2.0项目数据集

本模型旨在为CS224n的学生在为默认最终项目建立待超越的基线时节省GPU时间。用于微调此模型的训练集与官方训练集相同；不过，评估和模型选择是使用官方开发集的大约一半（随机选取的6078个示例）进行的。数据文件可在https://github.com/elgeish/squad/tree/master/data找到 — 这是2020年冬季版本。鉴于官方SQuAD2.0开发集包含该项目的测试集，学生必须确保不以任何方式使用官方SQuAD2.0开发集 — 包括使用在官方SQuAD2.0上微调的模型，因为它们在模型选择时使用了官方SQuAD2.0开发集。

✨ 主要特性

📊 结果

{
  "exact": 78.94044093451794,
  "f1": 81.7724930324639,
  "total": 6078,
  "HasAns_exact": 76.28865979381443,
  "HasAns_f1": 82.20385314478195,
  "HasAns_total": 2910,
  "NoAns_exact": 81.37626262626263,
  "NoAns_f1": 81.37626262626263,
  "NoAns_total": 3168,
  "best_exact": 78.95689371503784,
  "best_exact_thresh": 0.0,
  "best_f1": 81.78894581298378,
  "best_f1_thresh": 0.0
}

⚙️ 显著参数

{
  "do_lower_case": true,
  "doc_stride": 128,
  "fp16": false,
  "fp16_opt_level": "O1",
  "gradient_accumulation_steps": 24,
  "learning_rate": 3e-05,
  "max_answer_length": 30,
  "max_grad_norm": 1,
  "max_query_length": 64,
  "max_seq_length": 384,
  "model_name_or_path": "albert-base-v2",
  "model_type": "albert",
  "num_train_epochs": 3,
  "per_gpu_train_batch_size": 8,
  "save_steps": 5000,
  "seed": 42,
  "train_batch_size": 8,
  "version_2_with_negative": true,
  "warmup_steps": 0,
  "weight_decay": 0
}

💻 环境设置

{
  "transformers": "2.5.1",
  "pytorch": "1.4.0=py3.6_cuda10.1.243_cudnn7.6.3_0",
  "python": "3.6.5=hc3d631a_2",
  "os": "Linux 4.15.0-1060-aws #62-Ubuntu SMP Tue Feb 11 21:23:22 UTC 2020 x86_64 x86_64 x86_64 GNU/Linux",
  "gpu": "Tesla V100-SXM2-16GB"
}

📚 详细文档

📝 引用方式

@misc{elgeish2020gestalt,
  title={Gestalt: a Stacking Ensemble for SQuAD2.0},
  author={Mohamed El-Geish},
  journal={arXiv e-prints},
  archivePrefix={arXiv},
  eprint={2004.07067},
  year={2020},
}

🌟 相关模型

• elgeish/cs224n-squad2.0-albert-large-v2
• elgeish/cs224n-squad2.0-albert-xxlarge-v1
• elgeish/cs224n-squad2.0-distilbert-base-uncased
• elgeish/cs224n-squad2.0-roberta-base