Skip to content

Conversation

@ElizaWszola
Copy link
Contributor

@ElizaWszola ElizaWszola commented Oct 31, 2025

Fix optional vs. bool type issue in test_fused_quant_layernorm.py. This allows int8 tests to run correctly.

TODO: fix unit test failures

ElizaWszola and others added 4 commits October 31, 2025 03:53
Signed-off-by: ElizaWszola <ewszola@redhat.com>
Signed-off-by: yewentao256 <zhyanwentao@126.com>
Signed-off-by: yewentao256 <zhyanwentao@126.com>
@yewentao256 yewentao256 marked this pull request as ready for review November 3, 2025 16:45
@yewentao256 yewentao256 added the ready ONLY add when PR is ready to merge/full CI is needed label Nov 3, 2025
Copy link
Member

@yewentao256 yewentao256 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

After we fix the scale_ub case in unit test, we now face three known issues.

  • precision error for int8
  • precision error for fp8
  • illegal memory access without setting scale_ub

And I fix all of them in my commits

reserve core]$ pytest test_fused_quant_layernorm.py -x
============= test session starts ==============
platform linux -- Python 3.12.11, pytest-8.4.2, pluggy-1.6.0
rootdir: /home/yewentao256/vllm-source
configfile: pyproject.toml
plugins: anyio-4.11.0
collected 576 items                            

test_fused_quant_layernorm.py .......... [  1%]
........................................ [  8%]
........................................ [ 15%]
........................................ [ 22%]
........................................ [ 29%]
........................................ [ 36%]
........................................ [ 43%]
........................................ [ 50%]
........................................ [ 57%]
........................................ [ 64%]
........................................ [ 71%]
........................................ [ 78%]
........................................ [ 85%]
........................................ [ 92%]
........................................ [ 98%]
......                                   [100%]

= 576 passed, 2 warnings in 160.03s (0:02:40) ==

Signed-off-by: ElizaWszola <ewszola@redhat.com>
ElizaWszola added a commit to neuralmagic/vllm that referenced this pull request Nov 4, 2025
…or int8

Signed-off-by: ElizaWszola <ewszola@redhat.com>
Signed-off-by: ElizaWszola <ewszola@redhat.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

ready ONLY add when PR is ready to merge/full CI is needed

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants