DISABLED test_run2run_determinism_model_name_DistillGPT2_training_or_inference_training_precision_bfloat16 (__main__.DeterministicTest)
- Dominant language
- Python
- Stars
- 103k
- Forks
- 29.5k
- PR merge metrics
- PR metrics pending
Description
Platforms: linux, slow
This test was disabled because it is failing in CI. See [recent examples](https://hud.pytorch.org/flakytest?name=test_run2run_determinism_model_name_DistillGPT2_training_or_inference_training_precision_bfloat16&suite=DeterministicTest&limit=100) and the most recent trunk [workflow logs](https://github.com/pytorch/pytorch/runs/101075513496).
Over the past 6 hours, it has been determined flaky in 4 workflow(s) with 4 failures and 4 successes.
**Debugging instructions (after clicking on the recent samples link):**
DO NOT ASSUME THINGS ARE OKAY IF THE CI IS GREEN. We now shield flaky tests from developers so CI will thus be green but it will be harder to parse the logs.
To find relevant log snippets:
1. Click on the workflow logs linked above
2. Click on the Test step of the job so that it is expanded. Otherwise, the grepping will not work.
3. Grep for `test_run2run_determinism_model_name_DistillGPT2_training_or_inference_training_precision_bfloat16`
4. There should be several instances run (as flaky tests are rerun in CI) from which you can study the logs.
Sample error message
```
Traceback (most recent call last):
File "/__w/pytorch/pytorch/test/inductor/test_deterministic.py", line 232, in test_run2run_determinism
self.assertTrue(
File "/usr/lib/python3.10/unittest/case.py", line 687, in assertTrue
raise self.failureException(msg)
AssertionError: False is not true
stdout: cuda train DistillGPT2
Load model outputs from /tmp/tmpzphjmmz_/saved.pkl to compare
fail_accuracy
, stderr:
loading model: 0it [00:00, ?it/s][transformers] `loss_type=None` was set in the config but it is unrecognized. Using the default loss: `ForCausalLMLoss`.
WARNING:transformers.modeling_utils:`loss_type=None` was set in the config but it is unrecognized. Using the default loss: `ForCausalLMLoss`.
loading model: 0it [00:01, ?it/s]
W0904 14:09:00.567000 110133 torch/_logging/_internal.py:1394] [9/0] Profiler record function will be ignored
To execute this test, run the following from the base repo dir:
PYTORCH_TEST_WITH_SLOW=1 PYTORCH_TEST_SKIP_FAST=1 python test/inductor/test_deterministic.py DeterministicTest.test_run2run_determinism_model_name_DistillGPT2_training_or_inference_training_precision_bfloat16
This message can be suppressed by setting PYTORCH_PRINT_REPRO_ON_FAILURE=0
```
Test file path: `inductor/test_deterministic.py`
For all disabled tests (by GitHub issue), see https://hud.pytorch.org/disabled.
cc @voznesenskym @penguinwu @EikanWang @jgong5 @Guobing-Chen @XiaobingSuper @zhuhaozhe @blzheng @wenzhe-nrv @jiayisunx @ipiszy @kadeng @muchulee8 @amjames @chauhang @aakhundov @coconutruben @jataylo
Contributor guide
Research direction
Start with test/inductor/test_deterministic.py and run the supplied PYTORCH_TEST_WITH_SLOW=1 command for the named DeterministicTest case. Inspect the linked workflow logs and recent flaky-test samples, grepping for the test name and comparing its repeated runs. Done means the source of the nondeterministic failure is addressed and the test can run reliably rather than remaining disabled.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- machine-learning, testing
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Active
- Clarity
- Mostly clear
- Newbie friendliness
- 45/100