{
  "id": 490970,
  "title": "Compare BirdCLEF 2023 vs 2024",
  "url": "/competitions/birdclef-2024/discussion/490970",
  "author_name": "",
  "post_date": "2024-04-04T04:56:58.614155600Z",
  "votes": 67,
  "comment_count": 14,
  "views": 0,
  "content": "<h3>Location:</h3>\n<ul>\n<li>2023:  <strong>Kenya</strong>.</li>\n<li>2024:  <strong>Global Biodiversity Hotspot in the Western Ghats</strong>.</li>\n</ul>\n<h3>Test Soundscapes:</h3>\n<ul>\n<li>2023:  <strong>200 recordings</strong>, 10 minutes each.</li>\n<li>2024:  <strong>1,100 recordings</strong>, <strong>4 minutes</strong> each (<strong>short recordings</strong>)</li>\n</ul>\n<h3>Unlabeled Soundscapes:</h3>\n<ul>\n<li>2023: No unlabeled soundscapes.</li>\n<li>2024: <strong>unlabeled soundscapes</strong> - (<strong>extra pseudo labeling</strong>)</li>\n</ul>\n<h3>Data Freq &amp; Meta</h3>\n<ul>\n<li>32 kHz downsampled both and same meta fields</li>\n</ul>\n<h3># of Birds:</h3>\n<ul>\n<li>2023: <strong>264</strong></li>\n<li>2024: <strong>182</strong></li>\n</ul>\n<h3>Overall Birds:</h3>\n<ul>\n<li>common birds (14): barswa,blakit1,categr,comsan,eaywag1,gargan,greegr,hoopoe,laudov1,litegr,litswi1,piekin1,rerswa1,woosan ( <strong>can be used for training</strong> )</li>\n<li>extra birds (168): asbfly,ashdro1,ashpri1,ashwoo2,asikoe2,asiope1,aspfly1,aspswi1,barfly1,bcnher,bkcbul1,bkrfla1,bkskit1,bkwsti,bladro1,blaeag1,blhori1,blnmon1,blrwar1,bncwoo3,brakit1,brasta1,brcful1,brfowl1,brnhao1,brnshr,brodro1,brwjac1,brwowl1,btbeat1,bwfshr1,chbeat1,cohcuc1,comfla1,comgre,comior1,comkin1,commoo3,commyn,compea,comros,comtai1,copbar1,crbsun2,cregos1,crfbar1,crseag1,dafbab1,darter2,emedov2,eucdov,eurbla2,eurcoo,forwag1,gloibi,goflea1,graher1,grbeat1,grecou1,grefla1,grehor1,grejun2,grenig1,grewar3,grnsan,grnwar1,grtdro1,gryfra,grynig2,grywag,gybpri1,gyhcaf1,heswoo1,houcro1,houspa,inbrob1,indpit1,indrob1,indrol2,indtit1,ingori1,inpher1,insbab1,insowl1,integr,isbduc1,jerbus2,junbab2,junmyn1,junowl1,kenplo1,kerlau2,labcro1,lblwar1,lesyel1,lewduc1,lirplo,litgre1,litspi1,lobsun2,maghor2,malpar1,maltro1,malwoo1,marsan,mawthr1,moipig1,nilfly2,niwpig1,nutman,orihob2,oripip1,pabflo1,paisto1,piebus1,placuc3,plaflo1,plapri1,plhpar1,pomgrp2,purher1,pursun3,pursun4,purswa3,putbab1,redspu1,revbul,rewbul,rewlap1,rocpig,rorpar,rossta2,rufbab3,ruftre2,rufwoo2,rutfly6,sbeowl1,scamin3,shikra1,smamin1,sohmyn1,spepic1,spodov,spoowl1,sqtbul1,stbkin1,sttwoo1,thbwar1,tibfly3,tilwar1,vefnut1,vehpar1,wbbfly1,wemhar1,whbbul2,whbsho3,whbtre1,whbwag1,whbwat1,whbwoo2,whcbar1,whiter2,whrmun,whtkin2,wynlau1,yebbab1,yebbul3,zitcis1</li>\n</ul>\n<blockquote>\n  <p>The main difference to me is that we have 2.2x less time to perform the inference this year than in the 2023 competition.<br>\n  This year: 4400 minutes to predict. Last year, 2000 minutes to predict. by <a href=\"https://www.kaggle.com/cpmpml\" target=\"_blank\">@cpmpml</a> </p>\n</blockquote>",
  "messages": [
    {
      "id": "2734348",
      "postDate": "04/04/2024 04:56:58",
      "content": "<h3>Location:</h3>\n<ul>\n<li>2023:  <strong>Kenya</strong>.</li>\n<li>2024:  <strong>Global Biodiversity Hotspot in the Western Ghats</strong>.</li>\n</ul>\n<h3>Test Soundscapes:</h3>\n<ul>\n<li>2023:  <strong>200 recordings</strong>, 10 minutes each.</li>\n<li>2024:  <strong>1,100 recordings</strong>, <strong>4 minutes</strong> each (<strong>short recordings</strong>)</li>\n</ul>\n<h3>Unlabeled Soundscapes:</h3>\n<ul>\n<li>2023: No unlabeled soundscapes.</li>\n<li>2024: <strong>unlabeled soundscapes</strong> - (<strong>extra pseudo labeling</strong>)</li>\n</ul>\n<h3>Data Freq &amp; Meta</h3>\n<ul>\n<li>32 kHz downsampled both and same meta fields</li>\n</ul>\n<h3># of Birds:</h3>\n<ul>\n<li>2023: <strong>264</strong></li>\n<li>2024: <strong>182</strong></li>\n</ul>\n<h3>Overall Birds:</h3>\n<ul>\n<li>common birds (14): barswa,blakit1,categr,comsan,eaywag1,gargan,greegr,hoopoe,laudov1,litegr,litswi1,piekin1,rerswa1,woosan ( <strong>can be used for training</strong> )</li>\n<li>extra birds (168): asbfly,ashdro1,ashpri1,ashwoo2,asikoe2,asiope1,aspfly1,aspswi1,barfly1,bcnher,bkcbul1,bkrfla1,bkskit1,bkwsti,bladro1,blaeag1,blhori1,blnmon1,blrwar1,bncwoo3,brakit1,brasta1,brcful1,brfowl1,brnhao1,brnshr,brodro1,brwjac1,brwowl1,btbeat1,bwfshr1,chbeat1,cohcuc1,comfla1,comgre,comior1,comkin1,commoo3,commyn,compea,comros,comtai1,copbar1,crbsun2,cregos1,crfbar1,crseag1,dafbab1,darter2,emedov2,eucdov,eurbla2,eurcoo,forwag1,gloibi,goflea1,graher1,grbeat1,grecou1,grefla1,grehor1,grejun2,grenig1,grewar3,grnsan,grnwar1,grtdro1,gryfra,grynig2,grywag,gybpri1,gyhcaf1,heswoo1,houcro1,houspa,inbrob1,indpit1,indrob1,indrol2,indtit1,ingori1,inpher1,insbab1,insowl1,integr,isbduc1,jerbus2,junbab2,junmyn1,junowl1,kenplo1,kerlau2,labcro1,lblwar1,lesyel1,lewduc1,lirplo,litgre1,litspi1,lobsun2,maghor2,malpar1,maltro1,malwoo1,marsan,mawthr1,moipig1,nilfly2,niwpig1,nutman,orihob2,oripip1,pabflo1,paisto1,piebus1,placuc3,plaflo1,plapri1,plhpar1,pomgrp2,purher1,pursun3,pursun4,purswa3,putbab1,redspu1,revbul,rewbul,rewlap1,rocpig,rorpar,rossta2,rufbab3,ruftre2,rufwoo2,rutfly6,sbeowl1,scamin3,shikra1,smamin1,sohmyn1,spepic1,spodov,spoowl1,sqtbul1,stbkin1,sttwoo1,thbwar1,tibfly3,tilwar1,vefnut1,vehpar1,wbbfly1,wemhar1,whbbul2,whbsho3,whbtre1,whbwag1,whbwat1,whbwoo2,whcbar1,whiter2,whrmun,whtkin2,wynlau1,yebbab1,yebbul3,zitcis1</li>\n</ul>\n<blockquote>\n  <p>The main difference to me is that we have 2.2x less time to perform the inference this year than in the 2023 competition.<br>\n  This year: 4400 minutes to predict. Last year, 2000 minutes to predict. by <a href=\"https://www.kaggle.com/cpmpml\" target=\"_blank\">@cpmpml</a> </p>\n</blockquote>",
      "rawMarkdown": "### Location:\n- 2023:  **Kenya**.\n- 2024:  **Global Biodiversity Hotspot in the Western Ghats**.\n\n### Test Soundscapes:\n- 2023:  **200 recordings**, 10 minutes each.\n- 2024:  **1,100 recordings**, **4 minutes** each (**short recordings**)\n\n### Unlabeled Soundscapes:\n- 2023: No unlabeled soundscapes.\n- 2024: **unlabeled soundscapes** - (**extra pseudo labeling**)\n\n### Data Freq & Meta\n- 32 kHz downsampled both and same meta fields\n\n### # of Birds:\n- 2023: **264**\n- 2024: **182**\n\n\n\n\n\n### Overall Birds:\n- common birds (14): barswa,blakit1,categr,comsan,eaywag1,gargan,greegr,hoopoe,laudov1,litegr,litswi1,piekin1,rerswa1,woosan ( **can be used for training** )\n- extra birds (168): asbfly,ashdro1,ashpri1,ashwoo2,asikoe2,asiope1,aspfly1,aspswi1,barfly1,bcnher,bkcbul1,bkrfla1,bkskit1,bkwsti,bladro1,blaeag1,blhori1,blnmon1,blrwar1,bncwoo3,brakit1,brasta1,brcful1,brfowl1,brnhao1,brnshr,brodro1,brwjac1,brwowl1,btbeat1,bwfshr1,chbeat1,cohcuc1,comfla1,comgre,comior1,comkin1,commoo3,commyn,compea,comros,comtai1,copbar1,crbsun2,cregos1,crfbar1,crseag1,dafbab1,darter2,emedov2,eucdov,eurbla2,eurcoo,forwag1,gloibi,goflea1,graher1,grbeat1,grecou1,grefla1,grehor1,grejun2,grenig1,grewar3,grnsan,grnwar1,grtdro1,gryfra,grynig2,grywag,gybpri1,gyhcaf1,heswoo1,houcro1,houspa,inbrob1,indpit1,indrob1,indrol2,indtit1,ingori1,inpher1,insbab1,insowl1,integr,isbduc1,jerbus2,junbab2,junmyn1,junowl1,kenplo1,kerlau2,labcro1,lblwar1,lesyel1,lewduc1,lirplo,litgre1,litspi1,lobsun2,maghor2,malpar1,maltro1,malwoo1,marsan,mawthr1,moipig1,nilfly2,niwpig1,nutman,orihob2,oripip1,pabflo1,paisto1,piebus1,placuc3,plaflo1,plapri1,plhpar1,pomgrp2,purher1,pursun3,pursun4,purswa3,putbab1,redspu1,revbul,rewbul,rewlap1,rocpig,rorpar,rossta2,rufbab3,ruftre2,rufwoo2,rutfly6,sbeowl1,scamin3,shikra1,smamin1,sohmyn1,spepic1,spodov,spoowl1,sqtbul1,stbkin1,sttwoo1,thbwar1,tibfly3,tilwar1,vefnut1,vehpar1,wbbfly1,wemhar1,whbbul2,whbsho3,whbtre1,whbwag1,whbwat1,whbwoo2,whcbar1,whiter2,whrmun,whtkin2,wynlau1,yebbab1,yebbul3,zitcis1\n\n\n> The main difference to me is that we have 2.2x less time to perform the inference this year than in the 2023 competition.\nThis year: 4400 minutes to predict. Last year, 2000 minutes to predict. by @cpmpml",
      "votes": null
    },
    {
      "id": "2737593",
      "postDate": "04/05/2024 20:19:40",
      "content": "<p>Thank you!!</p>",
      "rawMarkdown": "Thank you!!",
      "votes": null
    },
    {
      "id": "2741090",
      "postDate": "04/08/2024 06:11:30",
      "content": "<p>That's great!</p>",
      "rawMarkdown": "That's great!",
      "votes": null
    },
    {
      "id": "2743177",
      "postDate": "04/09/2024 09:06:35",
      "content": "<p>Interesting!</p>",
      "rawMarkdown": "Interesting!",
      "votes": null
    },
    {
      "id": "2747028",
      "postDate": "04/11/2024 15:59:36",
      "content": "<p>It seems like in 2023 and previous years most of the solutions used a form of multilabel loss like focal. However, in 2024 people (myself included) are using multiclass loss. Is there something I'm missing that makes 2024 more amendable to multiclass classification?</p>",
      "rawMarkdown": "It seems like in 2023 and previous years most of the solutions used a form of multilabel loss like focal. However, in 2024 people (myself included) are using multiclass loss. Is there something I'm missing that makes 2024 more amendable to multiclass classification?",
      "votes": null
    },
    {
      "id": "2752468",
      "postDate": "04/15/2024 01:44:13",
      "content": "<p>Great report, thank you!</p>",
      "rawMarkdown": "Great report, thank you!",
      "votes": null
    },
    {
      "id": "2771201",
      "postDate": "04/24/2024 06:33:27",
      "content": "<p>thank you!!!</p>",
      "rawMarkdown": "thank you!!!",
      "votes": null
    },
    {
      "id": "2790682",
      "postDate": "05/03/2024 09:46:44",
      "content": "<p>Thanks for the comparison! 👍</p>\n<p>Just to let you know, you have a wrong formatting here:</p>\n<pre><code>\n2023: 264\n 2024: 182\n</code></pre>",
      "rawMarkdown": "Thanks for the comparison! 👍\n\nJust to let you know, you have a wrong formatting here:\n\n```markdown\n# of Birds:\n2023: 264\n- 2024: 182\n```",
      "votes": null
    },
    {
      "id": "2790710",
      "postDate": "05/03/2024 09:56:37",
      "content": "<p>Fixed <a href=\"https://www.kaggle.com/yassinealouini\" target=\"_blank\">@yassinealouini</a> Thank you</p>",
      "rawMarkdown": "Fixed @yassinealouini Thank you",
      "votes": null
    },
    {
      "id": "2790737",
      "postDate": "05/03/2024 10:02:19",
      "content": "<p>The main difference to me is that we have 2.2x less time to perform the inference this year than in the 2023 competition.</p>\n<p>This year: 4400 minutes to predict. Last year, 2000 minutes to predict.</p>",
      "rawMarkdown": "The main difference to me is that we have 2.2x less time to perform the inference this year than in the 2023 competition.\n\nThis year: 4400 minutes to predict. Last year, 2000 minutes to predict.",
      "votes": null
    },
    {
      "id": "2790788",
      "postDate": "05/03/2024 10:36:47",
      "content": "<p>Updated the topic <a href=\"https://www.kaggle.com/cpmpml\" target=\"_blank\">@cpmpml</a> </p>",
      "rawMarkdown": "Updated the topic @cpmpml",
      "votes": null
    },
    {
      "id": "2791112",
      "postDate": "05/03/2024 13:33:56",
      "content": "<p>For those that didn't understand this directly: </p>\n<ul>\n<li>We have about <strong>120 minutes inference CPU time</strong> to do all the predictions</li>\n<li>The total length of files (i.e. sum of all audio files) to predict is 4400 minutes this year and 2000 minutes in 2023</li>\n</ul>\n<p>I hope this helps those that didn't get it quickly.</p>",
      "rawMarkdown": "For those that didn't understand this directly: \n\n- We have about **120 minutes inference CPU time** to do all the predictions\n- The total length of files (i.e. sum of all audio files) to predict is 4400 minutes this year and 2000 minutes in 2023\n\nI hope this helps those that didn't get it quickly.",
      "votes": null
    },
    {
      "id": "2820895",
      "postDate": "05/17/2024 18:48:06",
      "content": "<p>Thank you very much this was helpful to me and my score increased by a bit because of this!</p>",
      "rawMarkdown": "Thank you very much this was helpful to me and my score increased by a bit because of this!",
      "votes": null
    },
    {
      "id": "2823580",
      "postDate": "05/19/2024 09:59:41",
      "content": "<p>Mucha gracias!<br>\nLocation:<br>\n2023: Kenya.<br>\n2024: India.</p>",
      "rawMarkdown": "Mucha gracias!\nLocation:\n2023: Kenya.\n2024: India.",
      "votes": null
    },
    {
      "id": "2856514",
      "postDate": "06/05/2024 10:33:41",
      "content": "<p>Thank you!</p>",
      "rawMarkdown": "Thank you!",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2737593,
      "author_name": "khoatran311",
      "author_url": "",
      "post_date": "04/05/2024 20:19:40",
      "content": "<p>Thank you!!</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2741090,
      "author_name": "sonialikhan",
      "author_url": "",
      "post_date": "04/08/2024 06:11:30",
      "content": "<p>That's great!</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2743177,
      "author_name": "guusbouwens",
      "author_url": "",
      "post_date": "04/09/2024 09:06:35",
      "content": "<p>Interesting!</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2747028,
      "author_name": "willrice",
      "author_url": "",
      "post_date": "04/11/2024 15:59:36",
      "content": "<p>It seems like in 2023 and previous years most of the solutions used a form of multilabel loss like focal. However, in 2024 people (myself included) are using multiclass loss. Is there something I'm missing that makes 2024 more amendable to multiclass classification?</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2752468,
      "author_name": "deepernet",
      "author_url": "",
      "post_date": "04/15/2024 01:44:13",
      "content": "<p>Great report, thank you!</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2771201,
      "author_name": "soratakaya",
      "author_url": "",
      "post_date": "04/24/2024 06:33:27",
      "content": "<p>thank you!!!</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2790682,
      "author_name": "yassinealouini",
      "author_url": "",
      "post_date": "05/03/2024 09:46:44",
      "content": "<p>Thanks for the comparison! 👍</p>\n<p>Just to let you know, you have a wrong formatting here:</p>\n<pre><code>\n2023: 264\n 2024: 182\n</code></pre>",
      "votes": null,
      "replies": [
        {
          "id": 2790710,
          "author_name": "seshurajup",
          "author_url": "",
          "post_date": "05/03/2024 09:56:37",
          "content": "<p>Fixed <a href=\"https://www.kaggle.com/yassinealouini\" target=\"_blank\">@yassinealouini</a> Thank you</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 2790737,
      "author_name": "cpmpml",
      "author_url": "",
      "post_date": "05/03/2024 10:02:19",
      "content": "<p>The main difference to me is that we have 2.2x less time to perform the inference this year than in the 2023 competition.</p>\n<p>This year: 4400 minutes to predict. Last year, 2000 minutes to predict.</p>",
      "votes": null,
      "replies": [
        {
          "id": 2790788,
          "author_name": "seshurajup",
          "author_url": "",
          "post_date": "05/03/2024 10:36:47",
          "content": "<p>Updated the topic <a href=\"https://www.kaggle.com/cpmpml\" target=\"_blank\">@cpmpml</a> </p>",
          "votes": null,
          "replies": [
            {
              "id": 2791112,
              "author_name": "yassinealouini",
              "author_url": "",
              "post_date": "05/03/2024 13:33:56",
              "content": "<p>For those that didn't understand this directly: </p>\n<ul>\n<li>We have about <strong>120 minutes inference CPU time</strong> to do all the predictions</li>\n<li>The total length of files (i.e. sum of all audio files) to predict is 4400 minutes this year and 2000 minutes in 2023</li>\n</ul>\n<p>I hope this helps those that didn't get it quickly.</p>",
              "votes": null,
              "replies": []
            }
          ]
        }
      ]
    },
    {
      "id": 2820895,
      "author_name": "max1mum",
      "author_url": "",
      "post_date": "05/17/2024 18:48:06",
      "content": "<p>Thank you very much this was helpful to me and my score increased by a bit because of this!</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2823580,
      "author_name": "hypnotu",
      "author_url": "",
      "post_date": "05/19/2024 09:59:41",
      "content": "<p>Mucha gracias!<br>\nLocation:<br>\n2023: Kenya.<br>\n2024: India.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 2856514,
      "author_name": "skanderkaroui",
      "author_url": "",
      "post_date": "06/05/2024 10:33:41",
      "content": "<p>Thank you!</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2734348": "### Location:\n- 2023:  **Kenya**.\n- 2024:  **Global Biodiversity Hotspot in the Western Ghats**.\n\n### Test Soundscapes:\n- 2023:  **200 recordings**, 10 minutes each.\n- 2024:  **1,100 recordings**, **4 minutes** each (**short recordings**)\n\n### Unlabeled Soundscapes:\n- 2023: No unlabeled soundscapes.\n- 2024: **unlabeled soundscapes** - (**extra pseudo labeling**)\n\n### Data Freq & Meta\n- 32 kHz downsampled both and same meta fields\n\n### # of Birds:\n- 2023: **264**\n- 2024: **182**\n\n\n\n\n\n### Overall Birds:\n- common birds (14): barswa,blakit1,categr,comsan,eaywag1,gargan,greegr,hoopoe,laudov1,litegr,litswi1,piekin1,rerswa1,woosan ( **can be used for training** )\n- extra birds (168): asbfly,ashdro1,ashpri1,ashwoo2,asikoe2,asiope1,aspfly1,aspswi1,barfly1,bcnher,bkcbul1,bkrfla1,bkskit1,bkwsti,bladro1,blaeag1,blhori1,blnmon1,blrwar1,bncwoo3,brakit1,brasta1,brcful1,brfowl1,brnhao1,brnshr,brodro1,brwjac1,brwowl1,btbeat1,bwfshr1,chbeat1,cohcuc1,comfla1,comgre,comior1,comkin1,commoo3,commyn,compea,comros,comtai1,copbar1,crbsun2,cregos1,crfbar1,crseag1,dafbab1,darter2,emedov2,eucdov,eurbla2,eurcoo,forwag1,gloibi,goflea1,graher1,grbeat1,grecou1,grefla1,grehor1,grejun2,grenig1,grewar3,grnsan,grnwar1,grtdro1,gryfra,grynig2,grywag,gybpri1,gyhcaf1,heswoo1,houcro1,houspa,inbrob1,indpit1,indrob1,indrol2,indtit1,ingori1,inpher1,insbab1,insowl1,integr,isbduc1,jerbus2,junbab2,junmyn1,junowl1,kenplo1,kerlau2,labcro1,lblwar1,lesyel1,lewduc1,lirplo,litgre1,litspi1,lobsun2,maghor2,malpar1,maltro1,malwoo1,marsan,mawthr1,moipig1,nilfly2,niwpig1,nutman,orihob2,oripip1,pabflo1,paisto1,piebus1,placuc3,plaflo1,plapri1,plhpar1,pomgrp2,purher1,pursun3,pursun4,purswa3,putbab1,redspu1,revbul,rewbul,rewlap1,rocpig,rorpar,rossta2,rufbab3,ruftre2,rufwoo2,rutfly6,sbeowl1,scamin3,shikra1,smamin1,sohmyn1,spepic1,spodov,spoowl1,sqtbul1,stbkin1,sttwoo1,thbwar1,tibfly3,tilwar1,vefnut1,vehpar1,wbbfly1,wemhar1,whbbul2,whbsho3,whbtre1,whbwag1,whbwat1,whbwoo2,whcbar1,whiter2,whrmun,whtkin2,wynlau1,yebbab1,yebbul3,zitcis1\n\n\n> The main difference to me is that we have 2.2x less time to perform the inference this year than in the 2023 competition.\nThis year: 4400 minutes to predict. Last year, 2000 minutes to predict. by @cpmpml",
    "2737593": "Thank you!!",
    "2741090": "That's great!",
    "2743177": "Interesting!",
    "2747028": "It seems like in 2023 and previous years most of the solutions used a form of multilabel loss like focal. However, in 2024 people (myself included) are using multiclass loss. Is there something I'm missing that makes 2024 more amendable to multiclass classification?",
    "2752468": "Great report, thank you!",
    "2771201": "thank you!!!",
    "2790682": "Thanks for the comparison! 👍\n\nJust to let you know, you have a wrong formatting here:\n\n```markdown\n# of Birds:\n2023: 264\n- 2024: 182\n```",
    "2790710": "Fixed @yassinealouini Thank you",
    "2790737": "The main difference to me is that we have 2.2x less time to perform the inference this year than in the 2023 competition.\n\nThis year: 4400 minutes to predict. Last year, 2000 minutes to predict.",
    "2790788": "Updated the topic @cpmpml",
    "2791112": "For those that didn't understand this directly: \n\n- We have about **120 minutes inference CPU time** to do all the predictions\n- The total length of files (i.e. sum of all audio files) to predict is 4400 minutes this year and 2000 minutes in 2023\n\nI hope this helps those that didn't get it quickly.",
    "2820895": "Thank you very much this was helpful to me and my score increased by a bit because of this!",
    "2823580": "Mucha gracias!\nLocation:\n2023: Kenya.\n2024: India.",
    "2856514": "Thank you!"
  },
  "source": "meta"
}