[OVEP] Enable stateful mode for Phi-silica models #821
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Kotomi-Du
commented
Oct 3, 2025
| if (gpu_or_npu) { | ||
| prefill_use_full_chat_history = true; | ||
| } | ||
| // bool gpu_or_npu = ((device.find("NPU") != std::string::npos) || (device.find("GPU") != std::string::npos)); |
There was a problem hiding this comment.
need to discuss with ORT-GenAI team how to handle this logic
Co-author: Beheshti, Nazanin
Kotomi-Du
force-pushed
the
make_stateful_phisilica
branch
from
0fe0302 to
1e132f3
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Description
Recognize other LLM models specifically for phi-silica models to trigger making stateful model.
Motivation and Context
Combined with #830 and #831 , these changes improved the memory footprint and performance for Phi-Silica workload. Without these changes, it consumed 16GB memory and got 1 fps when running the workload on OVEP GPU backend. After the change, the memory usage reduced to 3.7GB and the performance achieves 16fps.
Open
Should OVEP add an API in provider option to let user decide if they want to make the model stateful? If so, we don't need to hardcode the input name for specific models.