{
  "id": 390328,
  "title": "How to ensure the inference time limit in advance?",
  "url": "/competitions/asl-signs/discussion/390328",
  "author_name": "",
  "post_date": "2023-02-25T05:43:40.726008300Z",
  "votes": 6,
  "comment_count": 1,
  "views": 0,
  "content": "<p>Hi everyone,</p>\n<p>I'm new to TensorFlow Lite. This competition has a constraint of less than 100 milliseconds of latency per video. Say I want to train a large model in PyTorch or any other framework and then convert it to TensorFlow Lite. Is there any way I can find out how large my original model can get, apart from just submitting it and seeing if it fits?</p>\n<p>Thanks!</p>",
  "messages": [
    {
      "id": "2158738",
      "postDate": "02/25/2023 05:43:40",
      "content": "<p>Hi everyone,</p>\n<p>I'm new to TensorFlow Lite. This competition has a constraint of less than 100 milliseconds of latency per video. Say I want to train a large model in PyTorch or any other framework and then convert it to TensorFlow Lite. Is there any way I can find out how large my original model can get, apart from just submitting it and seeing if it fits?</p>\n<p>Thanks!</p>",
      "rawMarkdown": "Hi everyone,\n\nI'm new to TensorFlow Lite. This competition has a constraint of less than 100 milliseconds of latency per video. Say I want to train a large model in PyTorch or any other framework and then convert it to TensorFlow Lite. Is there any way I can find out how large my original model can get, apart from just submitting it and seeing if it fits?\n\nThanks!",
      "votes": null
    },
    {
      "id": "2175552",
      "postDate": "03/10/2023 00:09:09",
      "content": "<p>Adding to this, what hardware is the target? <br>\nLatency would depend on this &gt; GPUs TPUs -&gt; any other sample reference </p>",
      "rawMarkdown": "Adding to this, what hardware is the target? \nLatency would depend on this > GPUs TPUs -> any other sample reference",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 2175552,
      "author_name": "ratnesh1729",
      "author_url": "",
      "post_date": "03/10/2023 00:09:09",
      "content": "<p>Adding to this, what hardware is the target? <br>\nLatency would depend on this &gt; GPUs TPUs -&gt; any other sample reference </p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "2158738": "Hi everyone,\n\nI'm new to TensorFlow Lite. This competition has a constraint of less than 100 milliseconds of latency per video. Say I want to train a large model in PyTorch or any other framework and then convert it to TensorFlow Lite. Is there any way I can find out how large my original model can get, apart from just submitting it and seeing if it fits?\n\nThanks!",
    "2175552": "Adding to this, what hardware is the target? \nLatency would depend on this > GPUs TPUs -> any other sample reference"
  },
  "source": "meta"
}