{
  "id": 122776,
  "title": "Advice to beginners starting out?",
  "url": "/competitions/tensorflow2-question-answering/discussion/122776",
  "author_name": "Hrishikesh Athalye",
  "post_date": "2019-12-22T19:55:13.515000",
  "votes": 0,
  "comment_count": 0,
  "views": 0,
  "content": "<p>I am a beginner in NLP and have just finished reading the BERT and BERT Baseline on NQ papers with average understanding of both. As per my understanding, the inference of the pretrained model implemented in the Baseline is the first thing to try out, which I have done so far with the basic benchmark accuracy. What kernels/approaches should I look at next? How do I start improving upon my model by modifying the baseline?</p>",
  "messages": [
    {
      "id": 700909,
      "postDate": "2019-12-22T19:55:13.517Z",
      "content": "<p>I am a beginner in NLP and have just finished reading the BERT and BERT Baseline on NQ papers with average understanding of both. As per my understanding, the inference of the pretrained model implemented in the Baseline is the first thing to try out, which I have done so far with the basic benchmark accuracy. What kernels/approaches should I look at next? How do I start improving upon my model by modifying the baseline?</p>",
      "rawMarkdown": "I am a beginner in NLP and have just finished reading the BERT and BERT Baseline on NQ papers with average understanding of both. As per my understanding, the inference of the pretrained model implemented in the Baseline is the first thing to try out, which I have done so far with the basic benchmark accuracy. What kernels/approaches should I look at next? How do I start improving upon my model by modifying the baseline?"
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "700909": "I am a beginner in NLP and have just finished reading the BERT and BERT Baseline on NQ papers with average understanding of both. As per my understanding, the inference of the pretrained model implemented in the Baseline is the first thing to try out, which I have done so far with the basic benchmark accuracy. What kernels/approaches should I look at next? How do I start improving upon my model by modifying the baseline?"
  }
}