debug: 174065 #3123
debug: 174065 #3123
Conversation
There was a problem hiding this comment.
Pull request overview
This PR changes test markers from skipif to xfail for Windows platform tests related to ticket CVS-174065. This allows tests to run on Windows but marks them as expected to fail, providing better visibility into the actual test behavior compared to skipping them entirely.
Key changes:
- Replaced
@pytest.mark.skipifwith@pytest.mark.xfailfor Windows-specific test conditions - Standardized the condition syntax to use
condition=(sys.platform == "win32")
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
|
add this pr to collect log, for CVS-174065's logs are no longer available. |
There was a problem hiding this comment.
Pull request overview
Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
| print(f"\n{'='*60}") | ||
| print(f"[DEBUG] Model: {gguf_model_id}") | ||
| print(f"[DEBUG] Prompt: {repr(prompt)}") | ||
| print(f"[DEBUG] Pipeline: {pipeline_type}") | ||
| print(f"[DEBUG] HF input_ids: {input_ids.tolist()}") | ||
| ov_tokenized = ov_pipe_gguf.get_tokenizer().encode(prompt) | ||
| print(f"[DEBUG] OV input_ids: {ov_tokenized.input_ids.data.tolist()}") | ||
| print(f"[DEBUG] HF output: {repr(res_string_input_1)}") | ||
| print(f"[DEBUG] OV output: {repr(res_string_input_2)}") | ||
| print(f"[DEBUG] Match: {res_string_input_1 == res_string_input_2}") | ||
| print(f"{'='*60}\n") |
There was a problem hiding this comment.
Debug print statements should use a proper logging framework instead of print(). Consider using Python's logging module or pytest's logging capabilities to make debug output configurable and avoid polluting test output in normal runs.
|
I have reproduced this qwen2.5-0.5b-instruct-q4_0.gguf locally. Logging of 4 errors: |
|
Added log printing and test failed test one by one via logging of first error: It shows that model_gguf1 is |
|
As tokenizer is ok, I replaced qwen2.5-0.5b-instruct-q4_0.gguf with qwen2.5-1.5b-instruct-q4_0.gguf for further investigation. FAILED tests\python_tests\test_gguf_reader.py::test_full_gguf_pipeline[model_gguf1-multiple_special_tokens-enable_save_ov_model=False-PipelineType.STATEFUL] - assert '\n <div cla...\n <h1' == 'çççççççççççç...ççççççççççççç' |
|
Debug for multiple_special_tokens with 1.5B model: logging: \n
\n \n <h1' [DEBUG] OV output: 'çççççççççççççççççççççççççççççç' [DEBUG] Match: False |
1 participant
Description
CVS-174065
add this pr to collect log, for 174065's logs are no longer available.