Optimize Lossguide#2883
Conversation
merge catboost to levachev/catboost
|
I hereby agree to the terms of the CLA available at: link. |
|
copy of #2835 which cannot be merged automatically because of sync engine update |
|
Shipped! |
|
I hereby agree to the terms of the CLA available at: link. |
|
Shipped! |
❌ Missing CLA from one or more contributorsThanks for your pull request! In order for us (YANDEX LLC) to accept patches and other contributions from you, See CONTRIBUTING.md for more info about Yandex Contributor License Agreement. The following contributors were found: (Only the first commit for a unique contributor is listed) |
|
I hereby agree to the terms of the CLA available at: link. |
|
I hereby agree to the terms of the CLA available at: https://yandex.ru/legal/cla/?lang=ru |
✅ All contributors are covered under a CLA with YandexSee CONTRIBUTING.md for more info about Yandex Contributor License Agreement. The following contributors were found: (Only the first commit for a unique contributor is listed) |
|
Shipped! |
|
@Evgueni-Petrov-aka-espetrov has triggered the import of a pull request, but it hasn't started. If you are a Yandex employee, you can view the log. |
|
Shipped! |
|
@Evgueni-Petrov-aka-espetrov has imported your pull request. If you are a Yandex employee, you can view this diff. |
|
✅ This pull request is being closed because it has been successfully merged into our internal monorepository. |
Optimizing Lossguide with the subtract trick. When testing, it was found that the result will be different(compared to the master version) if you do not set the parameters of CatBoostRegressor random_strength=0, bootstrap_type="No", has_time=True Otherwise, the result will be different due to the fact that Ctx->LearnProgress->Rand.GenRand() will be different when splitting the leaf --- Co-authored-by: Levachev [[email protected]] Pull Request resolved: #2883 Co-authored-by: espetrov <[email protected]> commit_hash:eb90d7e7d00a4e328bd770507c73a7e1f305145e
2 participants
Optimizing Lossguide with the subtract trick.
When testing, it was found that the result will be different(compared to the master version) if you do not set the parameters of CatBoostRegressor
random_strength=0,
bootstrap_type="No",
has_time=True
Otherwise, the result will be different due to the fact that Ctx->LearnProgress->Rand.GenRand() will be different when splitting the leaf