Conversation
dkalinowski
force-pushed
the
npu_docs
branch
from
b242b08 to
82faddc
Compare
demos/embeddings/README.md
Outdated
There was a problem hiding this comment.
this is inconsistent with the command used later. should be /dev/accel
demos/embeddings/README.md
Outdated
There was a problem hiding this comment.
Only NPU tests are limited to qwen3-embeddings
There was a problem hiding this comment.
Yes and it is clearly stated in documentation. Scroll up to see full list of validated models on CPU/GPU. The list you comment clearly specifies models validated on NPU, which is correct
demos/embeddings/README.md
Outdated
There was a problem hiding this comment.
client code should be the same for all target devices
Co-authored-by: Trawinski, Dariusz <dariusz.trawinski@intel.com>
demos/embeddings/README.md
Outdated
There was a problem hiding this comment.
why do we have a separate table for models tested and tested on npu? It would be clearer to add a column with a checkbox for npu enabled models.
demos/embeddings/README.md
Outdated
There was a problem hiding this comment.
This should be in a the section related to model export above. No need to make it a separate chapter.
demos/embeddings/README.md
Outdated
There was a problem hiding this comment.
@rasapala @michalkulakowski do we have all tests completed to confirm int4 gives good results?
3 participants
🛠 Summary
Updating export and NPU usage.
🧪 Checklist
``