{
  "id": 441620,
  "title": "What Language model are you using ?",
  "url": "/competitions/bengaliai-speech/discussion/441620",
  "author_name": "",
  "post_date": "2023-09-19T13:54:18.102242Z",
  "votes": 1,
  "comment_count": 4,
  "views": 0,
  "content": "<p>I am currently using the Language model from the following link<br>\n<a href=\"https://huggingface.co/arijitx/wav2vec2-xls-r-300m-bengali\" target=\"_blank\">https://huggingface.co/arijitx/wav2vec2-xls-r-300m-bengali</a></p>\n<p>I am planning to build my own n-gram language model. <br>\nI would love to hear from others what type of model they are using.</p>\n<p>My current LB score is from fine-tuning the following model for 5 epochs<br>\n<a href=\"https://huggingface.co/ai4bharat/indicwav2vec_v1_bengali\" target=\"_blank\">https://huggingface.co/ai4bharat/indicwav2vec_v1_bengali</a></p>",
  "messages": [
    {
      "id": "2446584",
      "postDate": "09/19/2023 13:54:18",
      "content": "<p>I am currently using the Language model from the following link<br>\n<a href=\"https://huggingface.co/arijitx/wav2vec2-xls-r-300m-bengali\" target=\"_blank\">https://huggingface.co/arijitx/wav2vec2-xls-r-300m-bengali</a></p>\n<p>I am planning to build my own n-gram language model. <br>\nI would love to hear from others what type of model they are using.</p>\n<p>My current LB score is from fine-tuning the following model for 5 epochs<br>\n<a href=\"https://huggingface.co/ai4bharat/indicwav2vec_v1_bengali\" target=\"_blank\">https://huggingface.co/ai4bharat/indicwav2vec_v1_bengali</a></p>",
      "rawMarkdown": "I am currently using the Language model from the following link\nhttps://huggingface.co/arijitx/wav2vec2-xls-r-300m-bengali\n\n\n\nI am planning to build my own n-gram language model. \nI would love to hear from others what type of model they are using.\n\n\n\nMy current LB score is from fine-tuning the following model for 5 epochs\nhttps://huggingface.co/ai4bharat/indicwav2vec_v1_bengali",
      "votes": null
    },
    {
      "id": "2446746",
      "postDate": "09/19/2023 15:30:42",
      "content": "<p>I built a 5-gram LM and it improved the LB score by 0.023. Did you fine-tune the model on the whole dataset?</p>",
      "rawMarkdown": "I built a 5-gram LM and it improved the LB score by 0.023. Did you fine-tune the model on the whole dataset?",
      "votes": null
    },
    {
      "id": "2446773",
      "postDate": "09/19/2023 15:48:53",
      "content": "<p><a href=\"https://www.kaggle.com/dhakshiin1601\" target=\"_blank\">@dhakshiin1601</a>, <a href=\"https://www.kaggle.com/mbmmurad\" target=\"_blank\">@mbmmurad</a> </p>\n<p>What percentage of the training dataset are you using? And what is the average training and inference time? </p>",
      "rawMarkdown": "dhakshiin1601, @mbmmurad \n\nWhat percentage of the training dataset are you using? And what is the average training and inference time?",
      "votes": null
    },
    {
      "id": "2447078",
      "postDate": "09/19/2023 19:52:50",
      "content": "<p>I am currently using dataset shared in the following link<br>\n<a href=\"https://www.kaggle.com/competitions/bengaliai-speech/discussion/435300\" target=\"_blank\">https://www.kaggle.com/competitions/bengaliai-speech/discussion/435300</a></p>\n<p>I am planning to clean the competition training dataset and add it to the above dataset in my future experiments.</p>",
      "rawMarkdown": "I am currently using dataset shared in the following link\nhttps://www.kaggle.com/competitions/bengaliai-speech/discussion/435300\n\nI am planning to clean the competition training dataset and add it to the above dataset in my future experiments.",
      "votes": null
    },
    {
      "id": "2447079",
      "postDate": "09/19/2023 19:53:32",
      "content": "<p><a href=\"https://www.kaggle.com/mbmmurad\" target=\"_blank\">@mbmmurad</a> </p>\n<p>Which dataset did you use for training LM </p>",
      "rawMarkdown": "mbmmurad \n\nWhich dataset did you use for training LM",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2446746,
      "author_name": "mbmmurad",
      "author_url": "",
      "post_date": "09/19/2023 15:30:42",
      "content": "<p>I built a 5-gram LM and it improved the LB score by 0.023. Did you fine-tune the model on the whole dataset?</p>",
      "votes": null,
      "replies": [
        {
          "id": 2446773,
          "author_name": "rahim3",
          "author_url": "",
          "post_date": "09/19/2023 15:48:53",
          "content": "<p><a href=\"https://www.kaggle.com/dhakshiin1601\" target=\"_blank\">@dhakshiin1601</a>, <a href=\"https://www.kaggle.com/mbmmurad\" target=\"_blank\">@mbmmurad</a> </p>\n<p>What percentage of the training dataset are you using? And what is the average training and inference time? </p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 2447078,
          "author_name": "dhakshiin1601",
          "author_url": "",
          "post_date": "09/19/2023 19:52:50",
          "content": "<p>I am currently using dataset shared in the following link<br>\n<a href=\"https://www.kaggle.com/competitions/bengaliai-speech/discussion/435300\" target=\"_blank\">https://www.kaggle.com/competitions/bengaliai-speech/discussion/435300</a></p>\n<p>I am planning to clean the competition training dataset and add it to the above dataset in my future experiments.</p>",
          "votes": null,
          "replies": [
            {
              "id": 2447079,
              "author_name": "dhakshiin1601",
              "author_url": "",
              "post_date": "09/19/2023 19:53:32",
              "content": "<p><a href=\"https://www.kaggle.com/mbmmurad\" target=\"_blank\">@mbmmurad</a> </p>\n<p>Which dataset did you use for training LM </p>",
              "votes": null,
              "replies": []
            }
          ]
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "2446584": "I am currently using the Language model from the following link\nhttps://huggingface.co/arijitx/wav2vec2-xls-r-300m-bengali\n\n\n\nI am planning to build my own n-gram language model. \nI would love to hear from others what type of model they are using.\n\n\n\nMy current LB score is from fine-tuning the following model for 5 epochs\nhttps://huggingface.co/ai4bharat/indicwav2vec_v1_bengali",
    "2446746": "I built a 5-gram LM and it improved the LB score by 0.023. Did you fine-tune the model on the whole dataset?",
    "2446773": "dhakshiin1601, @mbmmurad \n\nWhat percentage of the training dataset are you using? And what is the average training and inference time?",
    "2447078": "I am currently using dataset shared in the following link\nhttps://www.kaggle.com/competitions/bengaliai-speech/discussion/435300\n\nI am planning to clean the competition training dataset and add it to the above dataset in my future experiments.",
    "2447079": "mbmmurad \n\nWhich dataset did you use for training LM"
  },
  "source": "meta"
}