kaivalnp commented on code in PR #13285:
URL: https://github.com/apache/lucene/pull/13285#discussion_r1599150976
##
lucene/core/src/java/org/apache/lucene/search/AbstractVectorSimilarityQuery.java:
##
@@ -105,13 +116,16 @@ public Scorer scorer(LeafReaderContext context) throws
kaivalnp commented on code in PR #13285:
URL: https://github.com/apache/lucene/pull/13285#discussion_r1599151876
##
lucene/core/src/java/org/apache/lucene/search/AbstractVectorSimilarityQuery.java:
##
@@ -144,22 +158,22 @@ protected boolean match(int doc) {
}
kaivalnp commented on code in PR #13285:
URL: https://github.com/apache/lucene/pull/13285#discussion_r1599150976
##
lucene/core/src/java/org/apache/lucene/search/AbstractVectorSimilarityQuery.java:
##
@@ -105,13 +116,16 @@ public Scorer scorer(LeafReaderContext context) throws
msokolov commented on code in PR #13285:
URL: https://github.com/apache/lucene/pull/13285#discussion_r1599136370
##
lucene/core/src/java/org/apache/lucene/search/AbstractVectorSimilarityQuery.java:
##
@@ -105,13 +116,16 @@ public Scorer scorer(LeafReaderContext context) throws
mikemccand commented on PR #13364:
URL: https://github.com/apache/lucene/pull/13364#issuecomment-2108796478
This is cool! In the hot case, do we expect `prefetch` to be a no-op? So
we are hoping for "first do no harm" in that case? (I haven't looked at
`MMapDirectory`'s impl yet). But i
mikemccand commented on code in PR #13364:
URL: https://github.com/apache/lucene/pull/13364#discussion_r1599078360
##
lucene/core/src/java/org/apache/lucene/codecs/lucene99/Lucene99PostingsReader.java:
##
@@ -1097,7 +1118,9 @@ public BlockImpactsDocsEnum(FieldInfo fieldInfo,
In
mikemccand commented on code in PR #13364:
URL: https://github.com/apache/lucene/pull/13364#discussion_r1599076213
##
lucene/core/src/java/org/apache/lucene/codecs/lucene99/Lucene99PostingsReader.java:
##
@@ -902,6 +917,12 @@ public int advance(int target) throws IOException {
mikemccand commented on code in PR #13364:
URL: https://github.com/apache/lucene/pull/13364#discussion_r1599076213
##
lucene/core/src/java/org/apache/lucene/codecs/lucene99/Lucene99PostingsReader.java:
##
@@ -902,6 +917,12 @@ public int advance(int target) throws IOException {
mikemccand commented on PR #13364:
URL: https://github.com/apache/lucene/pull/13364#issuecomment-2108781387
We might (eventually, later) consider an API change when pulling postings
that expresses that the caller intends to advance? Or, maybe simpler would be
expressing that caller will no
benwtrent merged PR #13365:
URL: https://github.com/apache/lucene/pull/13365
--
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.a
benwtrent commented on issue #12313:
URL: https://github.com/apache/lucene/issues/12313#issuecomment-2108168850
> Not sure I understand option 3. Are you thinking that graph has different
types of edges b/w documents based on diff. similarity functions? So if you
were using max similarity y
vigyasharma commented on issue #12313:
URL: https://github.com/apache/lucene/issues/12313#issuecomment-2108154488
@benwtrent I like the idea of having documents be vertices in the graph,
with an API that let's you iterate/access the different vectors per doc. It
would have an indexing time
yupeng9 commented on issue #13003:
URL: https://github.com/apache/lucene/issues/13003#issuecomment-2108153649
This is very interesting work. We saw Milvus published article on how GPU
accelerates vector search, which looks like a game changer.
```
For a batch size of 1, the T4 is 6.4x
rmuir commented on PR #13363:
URL: https://github.com/apache/lucene/pull/13363#issuecomment-2108089293
I like this much better from api perspective: it closer maps to `madvise()`
and to me is more straightforward. Especially as the current PRs out there to
use it (terms and postings) are us
benwtrent commented on code in PR #13365:
URL: https://github.com/apache/lucene/pull/13365#discussion_r1598645109
##
lucene/core/src/java/org/apache/lucene/search/FloatVectorSimilarityValuesSource.java:
##
@@ -36,29 +36,6 @@ public FloatVectorSimilarityValuesSource(float[] vecto
benwtrent commented on code in PR #13365:
URL: https://github.com/apache/lucene/pull/13365#discussion_r1598625772
##
lucene/core/src/java/org/apache/lucene/util/quantization/QuantizedByteVectorValues.java:
##
@@ -53,5 +53,5 @@ public final long cost() {
* @param query the qu
benwtrent merged PR #13356:
URL: https://github.com/apache/lucene/pull/13356
--
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.a
benwtrent commented on code in PR #13365:
URL: https://github.com/apache/lucene/pull/13365#discussion_r1598534710
##
lucene/core/src/java/org/apache/lucene/util/quantization/QuantizedByteVectorValues.java:
##
@@ -53,5 +53,5 @@ public final long cost() {
* @param query the qu
benwtrent commented on issue #12313:
URL: https://github.com/apache/lucene/issues/12313#issuecomment-2107660631
@vigyasharma @krickert There are a couple of ways to implement this natively
in Lucene.
1. Have each individual vector be a connection in the graph with some
resolution ba
benwtrent commented on code in PR #13356:
URL: https://github.com/apache/lucene/pull/13356#discussion_r1598523684
##
lucene/CHANGES.txt:
##
@@ -359,6 +359,8 @@ Bug Fixes
* GITHUB#12966: Aggregation facets no longer assume that aggregation values
are positive. (Stefan Vodita)
bruno-roustant merged PR #13327:
URL: https://github.com/apache/lucene/pull/13327
--
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...@luc
rmuir commented on PR #13364:
URL: https://github.com/apache/lucene/pull/13364#issuecomment-2107576749
Can we do better than blindly prefetching skipdata? Currently, skipdata is
not used until we advance() past doc in the first block:
https://github.com/apache/lucene/blob/83db8dfba3
vigyasharma commented on issue #12313:
URL: https://github.com/apache/lucene/issues/12313#issuecomment-2107568883
> In another scenario, the results would just return the top doc and not
repeat it.
I believe this is what the parent-block join implementation for vector
values does cur
ChrisHegarty commented on code in PR #13365:
URL: https://github.com/apache/lucene/pull/13365#discussion_r1598460737
##
lucene/core/src/java/org/apache/lucene/util/quantization/QuantizedByteVectorValues.java:
##
@@ -53,5 +53,5 @@ public final long cost() {
* @param query the
benwtrent opened a new pull request, #13365:
URL: https://github.com/apache/lucene/pull/13365
Follow up to: https://github.com/apache/lucene/pull/13181
I noticed the quantized interface had a slightly different name.
Additionally, testing showed we are inconsistent when there ar
jpountz opened a new pull request, #13364:
URL: https://github.com/apache/lucene/pull/13364
This uses the `IndexInput#prefetch` API for postings. This relies on
heuristics, as we don't know ahead of time what data we will need from a
postings list:
- Postings lists are prefetched entire
jpountz commented on code in PR #13359:
URL: https://github.com/apache/lucene/pull/13359#discussion_r1598128358
##
lucene/core/src/java/org/apache/lucene/codecs/lucene90/blocktree/SegmentTermsEnum.java:
##
@@ -307,6 +309,31 @@ private boolean setEOF() {
return true;
}
kaivalnp commented on PR #13285:
URL: https://github.com/apache/lucene/pull/13285#issuecomment-2107012946
Hi @benwtrent @vigyasharma could you help review this? Thanks!
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use t
28 matches
Mail list logo
