iverase commented on code in PR #14626:
URL: https://github.com/apache/lucene/pull/14626#discussion_r2103918228
##
lucene/core/src/java/org/apache/lucene/search/PointRangeQuery.java:
##
@@ -154,16 +154,18 @@ private Relation relate(byte[] minPackedValue, byte[]
maxPackedValue)
iverase commented on code in PR #14626:
URL: https://github.com/apache/lucene/pull/14626#discussion_r2103918228
##
lucene/core/src/java/org/apache/lucene/search/PointRangeQuery.java:
##
@@ -154,16 +154,18 @@ private Relation relate(byte[] minPackedValue, byte[]
maxPackedValue)
dungba88 commented on PR #14009:
URL: https://github.com/apache/lucene/pull/14009#issuecomment-2903403699
Let me rephrase it better. I'm not worrying about the computation cost, as
you mentioned they are the same.
In Amazon (where both Vigya and I are working at), we usually run
KnnF
dungba88 commented on code in PR #14009:
URL: https://github.com/apache/lucene/pull/14009#discussion_r2103857961
##
lucene/core/src/java/org/apache/lucene/search/RerankFloatVectorQuery.java:
##
@@ -0,0 +1,116 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
vigyasharma commented on PR #14009:
URL: https://github.com/apache/lucene/pull/14009#issuecomment-2903244426
> For the alternative implementation we leave that up to the final hit
collection to return the top-k based on a ranking mechanism
Those final TopDocs collectors will use the s
vigyasharma commented on code in PR #14009:
URL: https://github.com/apache/lucene/pull/14009#discussion_r2103784830
##
lucene/core/src/java/org/apache/lucene/search/RerankFloatVectorQuery.java:
##
@@ -0,0 +1,116 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under o
dungba88 commented on issue #14643:
URL: https://github.com/apache/lucene/issues/14643#issuecomment-2903095335
Not sure if this is relevant:
https://github.com/mikemccand/luceneutil/issues/385, but if it's the same
problem it will be fixed with latest luceneutil.
--
This is an automated
dungba88 commented on PR #14009:
URL: https://github.com/apache/lucene/pull/14009#issuecomment-2903029304
> This gets handled by the k you pass to searcher.search(), and the top docs
collector you use. It's same as any other query.
I guess the main difference is when the results got t
github-actions[bot] commented on PR #14626:
URL: https://github.com/apache/lucene/pull/14626#issuecomment-2902936990
This PR has not had activity in the past 2 weeks, labeling it as stale. If
the PR is waiting for review, notify the d...@lucene.apache.org list. Thank you
for your contributi
msokolov commented on PR #14009:
URL: https://github.com/apache/lucene/pull/14009#issuecomment-2902557094
Is it this?
https://github.com/apache/lucene/blob/main/lucene/queries/src/java/org/apache/lucene/queries/function/FunctionScoreQuery.java
--
This is an automated message from the
vigyasharma commented on PR #14009:
URL: https://github.com/apache/lucene/pull/14009#issuecomment-2902502742
> Lucene typically rewrites the Query first, and then createWeight on the
simplest form.
That's a good callout. I'll override the `rewrite` method to rewrite the
inner query.
msokolov commented on PR #14009:
URL: https://github.com/apache/lucene/pull/14009#issuecomment-2902339944
I haven't dug in deep yet, but along the lines of further generality, would
it make sense to accept a DoubleValuesSource (that can compute dot products
against a vector field) rather t
heemin32 commented on PR #13525:
URL: https://github.com/apache/lucene/pull/13525#issuecomment-2902300025
On a second thought, if user just diable indexing, implementing multi vector
in the same field might be same as having a separate field without indexing
capability.
--
This is an aut
msokolov commented on issue #14643:
URL: https://github.com/apache/lucene/issues/14643#issuecomment-2902293560
> Hmm it's odd for the 500K docs case that recall is so much better with
FEWER segments: .969 with 8 segments, .705 with 89 segments.
--
This is an automated message from the Apa
msokolov commented on issue #14681:
URL: https://github.com/apache/lucene/issues/14681#issuecomment-2902282294
This seems worth exploring. One question I have, after reading the linked
discussion thread about integrating JVector as an OpenSearch plugin, is about
this claim ("JVector is a th
dweiss merged PR #14703:
URL: https://github.com/apache/lucene/pull/14703
--
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: issues-unsubscr...@lucene.apac
heemin32 commented on PR #13525:
URL: https://github.com/apache/lucene/pull/13525#issuecomment-2902248141
I believe this proposal to add a new field for multi-vector support is
facing significant challenges primarily because we aim to support HNSW-based
search on it. However, if our goal we
heemin32 commented on PR #13525:
URL: https://github.com/apache/lucene/pull/13525#issuecomment-2902247646
Certainly! Here's a clearer and more polished rephrasing of your message:
---
I believe this proposal to add a new field for multi-vector support is
facing significant chal
github-actions[bot] commented on PR #13948:
URL: https://github.com/apache/lucene/pull/13948#issuecomment-2902069576
This PR does not have an entry in lucene/CHANGES.txt. Consider adding one.
If the PR doesn't need a changelog entry, then add the skip-changelog-check
label to it and you wil
Trey314159 commented on issue #14659:
URL: https://github.com/apache/lucene/issues/14659#issuecomment-2902054358
@praveen-d291: Thanks for the pull request! I was unsure how best to modify
the tests since I don't read Telugu. I couldn't tell what would make
natural-looking examples and I di
jpountz commented on code in PR #14702:
URL: https://github.com/apache/lucene/pull/14702#discussion_r2102989752
##
lucene/core/src/java/org/apache/lucene/codecs/KnnVectorsReader.java:
##
@@ -124,7 +124,7 @@ public abstract void search(
*
* The default implementation retu
msokolov merged PR #14696:
URL: https://github.com/apache/lucene/pull/14696
--
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: issues-unsubscr...@lucene.ap
ChrisHegarty merged PR #14700:
URL: https://github.com/apache/lucene/pull/14700
--
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: issues-unsubscr...@lucen
phb-ig commented on code in PR #14697:
URL: https://github.com/apache/lucene/pull/14697#discussion_r2102606331
##
lucene/queryparser/src/test/org/apache/lucene/queryparser/flexible/standard/nodes/TestAbstractRangeQueryNode.java:
##
@@ -0,0 +1,69 @@
+/*
+ * Licensed to the Apache
phb-ig commented on code in PR #14697:
URL: https://github.com/apache/lucene/pull/14697#discussion_r2102594352
##
lucene/CHANGES.txt:
##
@@ -854,7 +862,7 @@ Improvements
* GITHUB#13285: Early terminate graph searches of
AbstractVectorSimilarityQuery to follow timeout set fro
phb-ig commented on PR #14697:
URL: https://github.com/apache/lucene/pull/14697#issuecomment-2901267587
@stefanvodita I am the one who originally submitted the patches all that
time ago because the work I was doing at the time required inspecting the query
string and altering, then re-parsi
dungba88 commented on PR #14009:
URL: https://github.com/apache/lucene/pull/14009#issuecomment-2901240220
I changed the query to be generic but keep other the same. (Didn't have
JDK24 yet so build would fail.)
--
This is an automated message from the Apache Git Service.
To respond to the
github-actions[bot] commented on PR #14009:
URL: https://github.com/apache/lucene/pull/14009#issuecomment-2901186441
This PR does not have an entry in lucene/CHANGES.txt. Consider adding one.
If the PR doesn't need a changelog entry, then add the skip-changelog-check
label to it and you wil
github-actions[bot] commented on PR #14703:
URL: https://github.com/apache/lucene/pull/14703#issuecomment-2901110046
This PR does not have an entry in lucene/CHANGES.txt. Consider adding one.
If the PR doesn't need a changelog entry, then add the skip-changelog-check
label to it and you wil
dungba88 commented on PR #14009:
URL: https://github.com/apache/lucene/pull/14009#issuecomment-2901114417
No worry, I think generalization makes sense to me too. And I like the idea
of moving this forward.
The use of `KnnFloatVectorQuery` in this implementation only adds a small
bene
dweiss opened a new pull request, #14703:
URL: https://github.com/apache/lucene/pull/14703
(no comment)
--
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-ma
github-actions[bot] commented on PR #14702:
URL: https://github.com/apache/lucene/pull/14702#issuecomment-2900966501
This PR does not have an entry in lucene/CHANGES.txt. Consider adding one.
If the PR doesn't need a changelog entry, then add the skip-changelog-check
label to it and you wil
thecoop opened a new pull request, #14702:
URL: https://github.com/apache/lucene/pull/14702
Update the `IOContext` on `IndexInput` implementations rather than the
`ReadAdvice`. This change means that `ReadAdvice` is now only used by
`MMapDirectory` and friends. Note that this is a refactori
jpountz commented on PR #14690:
URL: https://github.com/apache/lucene/pull/14690#issuecomment-2900934920
I'm superseding this change with a more general one for now, which doesn't
introduce new public APIs: #14701. We can look into taking ideas from this PR
as follow-ups.
--
This is an a
jpountz closed pull request #14690: Speed up conjunctive queries that need
scores.
URL: https://github.com/apache/lucene/pull/14690
--
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 comme
jpountz commented on PR #14701:
URL: https://github.com/apache/lucene/pull/14701#issuecomment-2900931305
Filtered queries get a slowdown, but some important queries get a big
speedup:
```
TaskQPS baseline StdDevQPS
my_modified_version StdDev
github-actions[bot] commented on PR #14701:
URL: https://github.com/apache/lucene/pull/14701#issuecomment-2900925674
This PR does not have an entry in lucene/CHANGES.txt. Consider adding one.
If the PR doesn't need a changelog entry, then add the skip-changelog-check
label to it and you wil
jpountz opened a new pull request, #14701:
URL: https://github.com/apache/lucene/pull/14701
`MaxScoreBulkScorer` and `BlockMaxConjunctionBulkScorer` currently evaluate
hits in a doc-at-a-time (DAAT) fashion, meaning that they they look at all
their clauses to find the next doc and so forth
rmuir commented on issue #14659:
URL: https://github.com/apache/lucene/issues/14659#issuecomment-2900766946
The filter here is working as documented: problem is, user didn't read the
documentation. Just don't use the filter if you dont want the transformations
that it does.
--
This is an
praveen-d291 opened a new pull request, #14699:
URL: https://github.com/apache/lucene/pull/14699
Fixes: #14659
Remove incorrect Telugu వు/మ conflation in Indic Normalization. They look
similar, but they are distinct with different meanings.
Currently "వు" is mapped to "మ" in
rmuir commented on issue #14659:
URL: https://github.com/apache/lucene/issues/14659#issuecomment-2900761266
> It's like conflating "rn" and "m" to merge burn/bum and corn/com. It could
happen when reading quickly or with poor handwriting, but it is not something
that should happen for searc
mikemccand commented on issue #14630:
URL: https://github.com/apache/lucene/issues/14630#issuecomment-2900807348
Hmm @yugushihuang (on my team (Amazon product search) team) found this is
another way to query the kernel (our Amazon Linux 2 boxes seem not to have
`/proc/config.gz`):
``
ChrisHegarty opened a new pull request, #14700:
URL: https://github.com/apache/lucene/pull/14700
In a similar way to how `IndexOrDocValuesQuery` may rewrite to a
`MatchAllDocsQuery`, if either of the queries rewrites to match no docs, then
return a `MatchNoDocsQuery`.
--
This is an autom
github-actions[bot] commented on PR #14700:
URL: https://github.com/apache/lucene/pull/14700#issuecomment-2900736002
This PR does not have an entry in lucene/CHANGES.txt. Consider adding one.
If the PR doesn't need a changelog entry, then add the skip-changelog-check
label to it and you wil
ChrisHegarty merged PR #14695:
URL: https://github.com/apache/lucene/pull/14695
--
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: issues-unsubscr...@lucen
jpountz merged PR #14679:
URL: https://github.com/apache/lucene/pull/14679
--
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: issues-unsubscr...@lucene.apa
github-actions[bot] commented on PR #14699:
URL: https://github.com/apache/lucene/pull/14699#issuecomment-2900515377
This PR does not have an entry in lucene/CHANGES.txt. Consider adding one.
If the PR doesn't need a changelog entry, then add the skip-changelog-check
label to it and you wil
vigyasharma commented on PR #14009:
URL: https://github.com/apache/lucene/pull/14009#issuecomment-2900200570
@dungba88 : I shared an [alternate
implementation](https://github.com/apache/lucene/pull/14698) for
RerankVectorQuery that I feel can generalize more broadly.
**Main Differen
vigyasharma opened a new pull request, #14698:
URL: https://github.com/apache/lucene/pull/14698
This PR presents an alternate to #14009 for `RerankVectorQuery`. It takes an
input query, a target vector and a field to use for vector values, and rescores
the output of wrapped query with full
49 matches
Mail list logo
