krickert commented on code in PR #1468:
URL: https://github.com/apache/opennlp/pull/1468#discussion_r4282602717
##########
opennlp-core/opennlp-ml/opennlp-dl/src/main/java/opennlp/dl/vectors/SentenceVectorsDL.java:
##########
@@ -223,9 +278,8 @@ public float[] embed(final CharSequence text) {
/**
* {@inheritDoc}
*
- * <p>The inputs are tokenized up front, grouped by tokenized length, and
each group runs
- * through the session once with shape {@code [group size, length]}. A batch
never pads, so
- * every row is computed from the tensors its single-input call would have
used.</p>
+ * <p>The inputs are grouped by tokenized length, never padded, and each
group runs in
+ * inferences of at most {@value #MAX_BATCH_TOKEN_POSITIONS} token
positions.</p>
Review Comment:
Added: the cap is 32 rows at the BERT length of 512.
--
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]
