{ "id": 177833, "title": "What is the baseline score for GCS access?", "url": "/competitions/lyft-motion-prediction-autonomous-vehicles/discussion/177833", "author_name": "", "post_date": "2020-08-27T14:02:35.935304700Z", "votes": 2, "comment_count": 4, "views": 0, "content": "

Is the baseline for GCS access Pytorch Baseline - Inference 134 Notebook?

", "messages": [ { "id": "987772", "postDate": "08/27/2020 14:02:35", "content": "

Is the baseline for GCS access Pytorch Baseline - Inference 134 Notebook?

", "rawMarkdown": "Is the baseline for GCS access Pytorch Baseline - Inference 134 Notebook?", "votes": null }, { "id": "987782", "postDate": "08/27/2020 14:11:20", "content": "

@ks2019 According to the description in competition. They have mentioned the host baseline score will be available on leaderboard. I don't see any of the hosts score yet. May be it will come soon. And 'Pytorch Baseline - Inference 134' is from one of the competitors named peter not from host. so 134 is not the baseline.

", "rawMarkdown": "ks2019 According to the description in competition. They have mentioned the host baseline score will be available on leaderboard. I don't see any of the hosts score yet. May be it will come soon. And 'Pytorch Baseline - Inference 134' is from one of the competitors named peter not from host. so 134 is not the baseline.", "votes": null }, { "id": "987891", "postDate": "08/27/2020 15:42:30", "content": "

Not published yet. WIll come soon.

", "rawMarkdown": "Not published yet. WIll come soon.", "votes": null }, { "id": "987913", "postDate": "08/27/2020 16:03:09", "content": "

That \"baseline\" kernel is largely based on this notebook in l5kit (not rendering properly in github ATM for me, was before, can view it on nbviewer if not).
\nI gather the kaggle kernel is assuming the baseline will be based off this. Though the kaggle one has some differences- resnet18 whereas it's resnet50 in l5kit and the kaggle one uses 10 history frames whereas l5kit has 0 (though this seems a bit odd so maybe still being finalised).

", "rawMarkdown": "That \"baseline\" kernel is largely based on this [notebook ](https://github.com/lyft/l5kit/blob/v1.0.6/examples/agent_motion_prediction/agent_motion_prediction.ipynb) in `l5kit` (not rendering properly in github ATM for me, was before, can view it on [nbviewer](https://nbviewer.jupyter.org/github/lyft/l5kit/blob/v1.0.6/examples/agent_motion_prediction/agent_motion_prediction.ipynb) if not).\nI gather the kaggle kernel is assuming the baseline will be based off this. Though the kaggle one has some differences- resnet18 whereas it's resnet50 in `l5kit` and the kaggle one uses 10 history frames whereas `l5kit` has 0 (though this seems a bit odd so maybe still being finalised).", "votes": null }, { "id": "988211", "postDate": "08/27/2020 22:01:51", "content": "

print(KaggleDatasets().get_gcs_path(\"lyft-motion-prediction-autonomous-vehicles\"))
\nIt shows GCS address

", "rawMarkdown": "print(KaggleDatasets().get_gcs_path(\"lyft-motion-prediction-autonomous-vehicles\"))\nIt shows GCS address", "votes": null } ], "comments": [ { "id": 987782, "author_name": "deepakrajpurushothaman", "author_url": "", "post_date": "08/27/2020 14:11:20", "content": "

@ks2019 According to the description in competition. They have mentioned the host baseline score will be available on leaderboard. I don't see any of the hosts score yet. May be it will come soon. And 'Pytorch Baseline - Inference 134' is from one of the competitors named peter not from host. so 134 is not the baseline.

", "votes": null, "replies": [ { "id": 987913, "author_name": "thomasbrandon", "author_url": "", "post_date": "08/27/2020 16:03:09", "content": "

That \"baseline\" kernel is largely based on this notebook in l5kit (not rendering properly in github ATM for me, was before, can view it on nbviewer if not).
\nI gather the kaggle kernel is assuming the baseline will be based off this. Though the kaggle one has some differences- resnet18 whereas it's resnet50 in l5kit and the kaggle one uses 10 history frames whereas l5kit has 0 (though this seems a bit odd so maybe still being finalised).

", "votes": null, "replies": [] } ] }, { "id": 987891, "author_name": "iglovikov", "author_url": "", "post_date": "08/27/2020 15:42:30", "content": "

Not published yet. WIll come soon.

", "votes": null, "replies": [] }, { "id": 988211, "author_name": "doanquanvietnamca", "author_url": "", "post_date": "08/27/2020 22:01:51", "content": "

print(KaggleDatasets().get_gcs_path(\"lyft-motion-prediction-autonomous-vehicles\"))
\nIt shows GCS address

", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "987772": "Is the baseline for GCS access Pytorch Baseline - Inference 134 Notebook?", "987782": "ks2019 According to the description in competition. They have mentioned the host baseline score will be available on leaderboard. I don't see any of the hosts score yet. May be it will come soon. And 'Pytorch Baseline - Inference 134' is from one of the competitors named peter not from host. so 134 is not the baseline.", "987891": "Not published yet. WIll come soon.", "987913": "That \"baseline\" kernel is largely based on this [notebook ](https://github.com/lyft/l5kit/blob/v1.0.6/examples/agent_motion_prediction/agent_motion_prediction.ipynb) in `l5kit` (not rendering properly in github ATM for me, was before, can view it on [nbviewer](https://nbviewer.jupyter.org/github/lyft/l5kit/blob/v1.0.6/examples/agent_motion_prediction/agent_motion_prediction.ipynb) if not).\nI gather the kaggle kernel is assuming the baseline will be based off this. Though the kaggle one has some differences- resnet18 whereas it's resnet50 in `l5kit` and the kaggle one uses 10 history frames whereas `l5kit` has 0 (though this seems a bit odd so maybe still being finalised).", "988211": "print(KaggleDatasets().get_gcs_path(\"lyft-motion-prediction-autonomous-vehicles\"))\nIt shows GCS address" }, "source": "meta" }