{
  "id": 69890,
  "title": "Looking for team",
  "url": "/competitions/inclusive-images-challenge/discussion/69890",
  "author_name": "",
  "post_date": "2018-10-28T11:58:30.951140800Z",
  "votes": null,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Hi, I know is late, but if you are a bit stuck in your score and want to test if an extra pair or eyes, hands and a bunch of GPUs can help you push your score... Im looking for a team, I offer:</p>\n\n<ul>\n<li>GPUs allowed by the Google cloud credits + some of my GPUs (2x1080 Ti + 2x2080 Ti) if other resources allow</li>\n<li>Some ideas to incorporate word vectors into the pipeline (e.g. based on zero-shot learning, e.g. <a href=\"https://arxiv.org/abs/1803.11320\">Transductive Unbiased Embedding for Zero-Shot Learning</a> ) </li>\n<li>Knowledge of keras, pytorch and fastai.</li>\n</ul>\n\n<p>Rgs - Andres</p>",
  "messages": [
    {
      "id": "411533",
      "postDate": "10/28/2018 11:58:30",
      "content": "<p>Hi, I know is late, but if you are a bit stuck in your score and want to test if an extra pair or eyes, hands and a bunch of GPUs can help you push your score... Im looking for a team, I offer:</p>\n\n<ul>\n<li>GPUs allowed by the Google cloud credits + some of my GPUs (2x1080 Ti + 2x2080 Ti) if other resources allow</li>\n<li>Some ideas to incorporate word vectors into the pipeline (e.g. based on zero-shot learning, e.g. <a href=\"https://arxiv.org/abs/1803.11320\">Transductive Unbiased Embedding for Zero-Shot Learning</a> ) </li>\n<li>Knowledge of keras, pytorch and fastai.</li>\n</ul>\n\n<p>Rgs - Andres</p>",
      "rawMarkdown": "Hi, I know is late, but if you are a bit stuck in your score and want to test if an extra pair or eyes, hands and a bunch of GPUs can help you push your score... Im looking for a team, I offer:\n\n - GPUs allowed by the Google cloud credits + some of my GPUs (2x1080 Ti + 2x2080 Ti) if other resources allow\n - Some ideas to incorporate word vectors into the pipeline (e.g. based on zero-shot learning, e.g. [Transductive Unbiased Embedding for Zero-Shot Learning][1] ) \n - Knowledge of keras, pytorch and fastai.\n\nRgs - Andres\n\n\n  [1]: https://arxiv.org/abs/1803.11320",
      "votes": null
    },
    {
      "id": "412323",
      "postDate": "10/30/2018 01:45:02",
      "content": "<p>As far as I know, kubernetes-gpu from NGC [k8s] allows to combine distributed GPU systems and aggregate their computational power. If you create  a distributed team and aggregate multiple GPUs - that will be a beneficial advantage, in my opinion.</p>",
      "rawMarkdown": "As far as I know, kubernetes-gpu from NGC [k8s] allows to combine distributed GPU systems and aggregate their computational power. If you create  a distributed team and aggregate multiple GPUs - that will be a beneficial advantage, in my opinion.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 412323,
      "author_name": "avolod",
      "author_url": "",
      "post_date": "10/30/2018 01:45:02",
      "content": "<p>As far as I know, kubernetes-gpu from NGC [k8s] allows to combine distributed GPU systems and aggregate their computational power. If you create  a distributed team and aggregate multiple GPUs - that will be a beneficial advantage, in my opinion.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "411533": "Hi, I know is late, but if you are a bit stuck in your score and want to test if an extra pair or eyes, hands and a bunch of GPUs can help you push your score... Im looking for a team, I offer:\n\n - GPUs allowed by the Google cloud credits + some of my GPUs (2x1080 Ti + 2x2080 Ti) if other resources allow\n - Some ideas to incorporate word vectors into the pipeline (e.g. based on zero-shot learning, e.g. [Transductive Unbiased Embedding for Zero-Shot Learning][1] ) \n - Knowledge of keras, pytorch and fastai.\n\nRgs - Andres\n\n\n  [1]: https://arxiv.org/abs/1803.11320",
    "412323": "As far as I know, kubernetes-gpu from NGC [k8s] allows to combine distributed GPU systems and aggregate their computational power. If you create  a distributed team and aggregate multiple GPUs - that will be a beneficial advantage, in my opinion."
  },
  "source": "meta"
}