[
https://issues.apache.org/jira/browse/SOLR-8492?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16102919#comment-16102919
]
Tommaso Teofili commented on SOLR-8492:
---
sure Joel, I'll have a look and let you know.
> Add
[
https://issues.apache.org/jira/browse/SOLR-8492?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15282757#comment-15282757
]
Joel Bernstein commented on SOLR-8492:
--
[~teofili], I'd like to get another set of eyes on this. I've
[
https://issues.apache.org/jira/browse/SOLR-8492?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15264975#comment-15264975
]
Joel Bernstein commented on SOLR-8492:
--
Yep, I see it. If the client cache wasn't set it was creating
[
https://issues.apache.org/jira/browse/SOLR-8492?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15264923#comment-15264923
]
Cao Manh Dat commented on SOLR-8492:
[~joel.bernstein] the mem leak appear in LogitCall class, whe
[
https://issues.apache.org/jira/browse/SOLR-8492?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15263956#comment-15263956
]
Joel Bernstein commented on SOLR-8492:
--
Where was the memory leak?
> Add LogisticRegressionQuery and
[
https://issues.apache.org/jira/browse/SOLR-8492?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15226306#comment-15226306
]
Alessandro Benedetti commented on SOLR-8492:
Will try to take a look this week and give a
[
https://issues.apache.org/jira/browse/SOLR-8492?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15226295#comment-15226295
]
Joel Bernstein commented on SOLR-8492:
--
I haven't yet had a chance to review [~caomanhdat]'s last
[
https://issues.apache.org/jira/browse/SOLR-8492?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15226032#comment-15226032
]
Alessandro Benedetti commented on SOLR-8492:
Guys,
great work !
I have been following the
[
https://issues.apache.org/jira/browse/SOLR-8492?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15154628#comment-15154628
]
Joel Bernstein commented on SOLR-8492:
--
[~caomanhdat], this sounds like a great patch. Having tests
[
https://issues.apache.org/jira/browse/SOLR-8492?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15152021#comment-15152021
]
Cao Manh Dat commented on SOLR-8492:
Right, It seem that we missing the bias coefficients. I will fix
[
https://issues.apache.org/jira/browse/SOLR-8492?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15146074#comment-15146074
]
Joel Bernstein commented on SOLR-8492:
--
[~caomanhdat], I've been testing out the algorithm by
[
https://issues.apache.org/jira/browse/SOLR-8492?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15093943#comment-15093943
]
Joel Bernstein commented on SOLR-8492:
--
I think this is an good idea when working with a cluster with
[
https://issues.apache.org/jira/browse/SOLR-8492?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15093927#comment-15093927
]
Joel Bernstein commented on SOLR-8492:
--
Ok, I think I see how the positive label is being used. So if
[
https://issues.apache.org/jira/browse/SOLR-8492?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15093168#comment-15093168
]
Cao Manh Dat commented on SOLR-8492:
Beautiful patch!
> Add LogisticRegressionQuery and LogitStream
>
[
https://issues.apache.org/jira/browse/SOLR-8492?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15091233#comment-15091233
]
Joel Bernstein commented on SOLR-8492:
--
[~caomanhdat], I was comparing the algorithm in the patch you
[
https://issues.apache.org/jira/browse/SOLR-8492?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15091290#comment-15091290
]
Cao Manh Dat commented on SOLR-8492:
He still use Gradient Descent
{code}
Gradient[i] =
[
https://issues.apache.org/jira/browse/SOLR-8492?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15090713#comment-15090713
]
Joel Bernstein commented on SOLR-8492:
--
[~caomanhdat], I pulled down your latest patch, read through
[
https://issues.apache.org/jira/browse/SOLR-8492?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15089149#comment-15089149
]
Joel Bernstein commented on SOLR-8492:
--
I should have a chance to review the latest patch on this
[
https://issues.apache.org/jira/browse/SOLR-8492?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15085529#comment-15085529
]
Joel Bernstein commented on SOLR-8492:
--
Great, I'll take a look.
> Add LogisticRegressionQuery and
[
https://issues.apache.org/jira/browse/SOLR-8492?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15085527#comment-15085527
]
Joel Bernstein commented on SOLR-8492:
--
[~caomanhdat], thanks for the comments!
I'll review the
[
https://issues.apache.org/jira/browse/SOLR-8492?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15085528#comment-15085528
]
Joel Bernstein commented on SOLR-8492:
--
Yes this is a problem I noticed after I put the patch up.
>
[
https://issues.apache.org/jira/browse/SOLR-8492?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15085585#comment-15085585
]
Cao Manh Dat commented on SOLR-8492:
As pointed out in this paper :
[
https://issues.apache.org/jira/browse/SOLR-8492?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15085671#comment-15085671
]
Joel Bernstein commented on SOLR-8492:
--
It is indeed very costly to add the randomization. Perhaps we
[
https://issues.apache.org/jira/browse/SOLR-8492?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15086500#comment-15086500
]
Cao Manh Dat commented on SOLR-8492:
Gradient Ascent find maximum for a function. Beside, the gradient
[
https://issues.apache.org/jira/browse/SOLR-8492?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15086453#comment-15086453
]
Joel Bernstein commented on SOLR-8492:
--
I think the difference might be I actually used a *Gradient
[
https://issues.apache.org/jira/browse/SOLR-8492?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15086693#comment-15086693
]
Joel Bernstein commented on SOLR-8492:
--
Ok, it sounds like I just had the wrong formula. I'll test out
[
https://issues.apache.org/jira/browse/SOLR-8492?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=15085146#comment-15085146
]
Cao Manh Dat commented on SOLR-8492:
What a wonderful patch. I'm very excited on implementing ml
27 matches
Mail list logo
