Make GPTQ test less flaky (#1295)

# What does this PR do?

<!--
Congratulations! You've made it this far! You're not quite done yet
though.

Once merged, your PR is going to appear in the release notes with the
title you set, so make sure it's a great title that fully reflects the
extent of your awesome contribution.

Then, please replace this with a description of the change and which
issue is fixed (if applicable). Please also include relevant motivation
and context. List any dependencies (if any) that are required for this
change.

Once you're done, someone will review your PR shortly (see the section
"Who can review?" below to tag some potential reviewers). They may
suggest changes to make the code even better. If no one reviewed your PR
after a week has passed, don't hesitate to post a new comment
@-mentioning the same persons---sometimes notifications get lost.
-->

<!-- Remove if not applicable -->

Fixes # (issue)


## Before submitting
- [ ] This PR fixes a typo or improves the docs (you can dismiss the
other checks if that's the case).
- [ ] Did you read the [contributor
guideline](https://github.com/huggingface/transformers/blob/main/CONTRIBUTING.md#start-contributing-pull-requests),
      Pull Request section?
- [ ] Was this discussed/approved via a Github issue or the
[forum](https://discuss.huggingface.co/)? Please add a link
      to it if that's the case.
- [ ] Did you make sure to update the documentation with your changes?
Here are the
[documentation
guidelines](https://github.com/huggingface/transformers/tree/main/docs),
and
[here are tips on formatting
docstrings](https://github.com/huggingface/transformers/tree/main/docs#writing-source-documentation).
- [ ] Did you write any new necessary tests?


## Who can review?

Anyone in the community is free to review the PR once the tests have
passed. Feel free to tag
members/contributors who may be interested in your PR.

<!-- Your PR will be replied to more quickly if you can figure out the
right person to tag with @


@OlivierDehaene OR @Narsil

 -->

integration-tests/conftest.py

@@ -24,6 +24,7 @@
 
 
 class ResponseComparator(JSONSnapshotExtension):
+    rtol = 0.2
     def serialize(
         self,
         data,
@@ -58,7 +59,7 @@ def eq_token(token: Token, other: Token) -> bool:
             return (
                 token.id == other.id
                 and token.text == other.text
-                and math.isclose(token.logprob, other.logprob, rel_tol=0.2)
+                and math.isclose(token.logprob, other.logprob, rel_tol=self.rtol)
                 and token.special == other.special
             )
 
@@ -68,7 +69,7 @@ def eq_prefill_token(prefill_token: InputToken, other: InputToken) -> bool:
                     prefill_token.id == other.id
                     and prefill_token.text == other.text
                     and (
-                        math.isclose(prefill_token.logprob, other.logprob, rel_tol=0.2)
+                        math.isclose(prefill_token.logprob, other.logprob, rel_tol=self.rtol)
                         if prefill_token.logprob is not None
                         else prefill_token.logprob == other.logprob
                     )
@@ -148,6 +149,10 @@ def eq_response(response: Response, other: Response) -> bool:
         )
 
 
+class GenerousResponseComparator(ResponseComparator):
+    # Needed for GPTQ with exllama which has serious numerical fluctuations.
+    rtol = 0.75
+
 class LauncherHandle:
     def __init__(self, port: int):
         self.client = AsyncClient(f"http://localhost:{port}")
@@ -193,6 +198,10 @@ def _inner_health(self) -> bool:
 def response_snapshot(snapshot):
     return snapshot.use_extension(ResponseComparator)
 
+@pytest.fixture
+def generous_response_snapshot(snapshot):
+    return snapshot.use_extension(GenerousResponseComparator)
+
 
 @pytest.fixture(scope="module")
 def event_loop():

integration-tests/models/test_flash_starcoder_gptq.py

@@ -15,20 +15,20 @@ async def flash_starcoder_gptq(flash_starcoder_gptq_handle):
 
 @pytest.mark.asyncio
 @pytest.mark.private
-async def test_flash_starcoder_gptq(flash_starcoder_gptq, response_snapshot):
+async def test_flash_starcoder_gptq(flash_starcoder_gptq, generous_response_snapshot):
     response = await flash_starcoder_gptq.generate(
         "def geometric_mean(L: List[float]):",
         max_new_tokens=20,
         decoder_input_details=True,
     )
     assert response.details.generated_tokens == 20
-    assert response == response_snapshot
+    assert response == generous_response_snapshot
 
 
 @pytest.mark.asyncio
 @pytest.mark.private
 async def test_flash_starcoder_gptq_default_params(
-    flash_starcoder_gptq, response_snapshot
+    flash_starcoder_gptq, generous_response_snapshot
 ):
     response = await flash_starcoder_gptq.generate(
         "def geometric_mean(L: List[float]):",
@@ -39,13 +39,13 @@ async def test_flash_starcoder_gptq_default_params(
         seed=0,
     )
     assert response.details.generated_tokens == 20
-    assert response == response_snapshot
+    assert response == generous_response_snapshot
 
 
 @pytest.mark.asyncio
 @pytest.mark.private
 async def test_flash_starcoder_gptq_load(
-    flash_starcoder_gptq, generate_load, response_snapshot
+    flash_starcoder_gptq, generate_load, generous_response_snapshot
 ):
     responses = await generate_load(
         flash_starcoder_gptq,
@@ -57,4 +57,4 @@ async def test_flash_starcoder_gptq_load(
     assert len(responses) == 4
     assert all([r.generated_text == responses[0].generated_text for r in responses])
 
-    assert responses == response_snapshot
+    assert responses == generous_response_snapshot