krickert commented on code in PR #1462:
URL: https://github.com/apache/opennlp/pull/1462#discussion_r4283462236
##########
opennlp-docs/src/docbkx/machine-learning.xml:
##########
@@ -183,5 +183,26 @@ try (TextEmbedder embedder =
providers.select(spec).create(spec)) {
has no provider until another one is installed.
Existing <code>SentenceVectorsDL</code>
constructors remain available.
</para>
+ <para>
+ To compute the vectors on a GPU, depend on
<code>opennlp-dl-gpu</code> in place of
+ <code>opennlp-dl</code> and pass
<code>InferenceOptions</code> to
+ <code>SentenceVectorsDL</code>.
<code>opennlp-dl-gpu</code> supports linux-x64 and
Review Comment:
I agree. I'm imagining a good collaborative blog post on how we can expand
the platform with multi-arch support.
I especially hope the Raspberry Pi angle attracts people. Too bad they cost
so much rn, if the price eases up this is a great low-cost student/hacker
learning opportunity. You can quickly tie language into a GPU and really
understand what it's doing without breaking the bank.
Since we didn't "officially" do the so much support publicly, I withheld it
from the docs but look forward to publishing a first round of tests in a post.
It'll make a great transition into the importance of addons, and makes a clear
statement (without being explicit) that addons are not second class citizens -
but real addons that allow for more experimentation and easy expansion of our
platform. Also, modern java doesn't need a bloated framework to be expansive -
base JDK does a great job of it now.
Finally, everyone seems to be all about embeddings these days, I hope our
platform helps get people closer to know what they actually are and do, not
just be a duct tape API to be a checkbox on someone's RAG checklist.
--
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]
