Study googlenet using PSGD with LMDB and hadoop parquet #146
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…20 epocs with seed intelligence around 2-5%. Data is partitioned into many lmdb without overlapping. Suggested study area. using Tensorflow as driver. Cause Caffe have a bug of auto switch from GPU to cpu with no reason. The best goal is the data distribution. Which I have not been able to dig out due to the caffe switch issue. Using predicative pushed down enabled spark sql should let you retrieving small part of data very efficiently... But this may not prove to be correct.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
No description provided.