Bimlesh759-AI edited a comment on issue #52:
URL: 
https://github.com/apache/incubator-bluemarlin/issues/52#issuecomment-1054318434


   1.
   **Details about training datasets:**
   Positive labels - 217949105
   Negative labels- 233029477
   
   total number of rows in datasets- 450978582
   total number of rows in datasets with unique aid_index (or aid)- 91152601
   
   Below table shows no. of rows (count) with particular keyword id.
   +-------+---------+
   |keyword|    count|
   +-------+---------+
   |     16| 13850541|
   |     17|  6065903|
   |      7| 10124720|
   |     14| 14468395|
   |     12|102332499|
   |      8| 14433430|
   |     10| 22725438|
   |     11|  7802683|
   |      2|  3771215|
   |      3|  9103754|
   |      5|  6555940|
   |     15| 18514390|
   |      9|  8764869|
   |      0| 12021109|
   |      1|127750503|
   |     13|  2956558|
   |      6| 52615187|
   |      4| 17121448|
   +-------+---------+
   
   **Details about test datasets:**
   Positive labels - 18131220
   Negative labels- 48307204
   
   total number of rows in datasets-66438424
   total number of rows in datasets with unique aid_index (or aid)- 30294628
   
   Below table shows no. of rows (count) with particular keyword id.
   +-------+--------+
   |keyword|   count|
   +-------+--------+
   |     16| 2127218|
   |     17| 1646675|
   |      7| 2001651|
   |     14| 2309676|
   |     12|11349194|
   |      8| 2383793|
   |     10| 3709643|
   |     11| 1878372|
   |      2|  949824|
   |      3| 1919426|
   |      5| 1417523|
   |     15| 2794550|
   |      9| 2006817|
   |      0| 2135362|
   |      1|16328372|
   |     13|  950384|
   |      6| 7033431|
   |      4| 3496513|
   +-------+--------+
   
   Above is the details regarding the training and test datasets. Please 
suggest if it is possible to reduce it without compromising on model accuracy 
in future.
   
   2. Currently we have not tried TF2.0 for multiple gpu  but implementation 
for multiple gpu is underway using TF1 +  horovod
   
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


Reply via email to