Update'unit_batch_dynamic_prepacked' tests to use ASSERT_NEAR instead of ASSERT_EQ (#157860) #157861
Conversation
psiddh
requested review from
digantdesai,
jerryzh168,
jianyuh,
kimishpatel and
salilsdesai
as code owners
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/157861
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit d4f38ab with merge base 86251ef ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
pytorch-bot
bot
added
module: cpu
|
This pull request was exported from Phabricator. Differential Revision: D77911682 |
|
Please update the summary. Original code wasnt checking for absolute equivalence. It was checking for 4 ULPs but didnt allow for any way change the ulp. Your change enables the vheck for 3 ulps |
… of ASSERT_EQ (pytorch#157861) Summary: Replaced ASSERT_FLOAT_EQ which defaults to fixed kMaxUlps ( = 4-ULP , See gtest-internal.h) with ASSERT_NEAR which lets us set epsilon to 1e-3, (approximately 3 ULPs). This allows for slightly stricter and tunable comparison. Test Plan: **Before Fix** ✗ Fail: qnnpack:pytorch_qnnpack_testApple - FULLY_CONNECTED_SPARSE_OP_8x1/unit_batch_dynamic_prepacked (0.0s) 'Expected equality of these values: output_dynamic[i * outputChannels() + c] Which is: 9.9160004 accumulators_float[i * outputChannels() + c] Which is: 9.9159956 at 0, 17: reference = 9.9159955978393555, optimized = 9.9160003662109375 ------------------------------ **After Fix** Everything passes Rollback Plan: Differential Revision: D77911682
|
This pull request was exported from Phabricator. Differential Revision: D77911682 |
psiddh
removed request for
jerryzh168,
jianyuh,
larryliu0820,
lucylq,
mergennachin,
metascroy and
salilsdesai
|
@pytorchbot merge (Initiating merge automatically since Phabricator Diff has merged) |
Merge startedYour change will be merged once all checks pass (ETA 0-4 Hours). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
i dont think anyone except some unittest. well i will have to eat my work, some model in some dark corner still might be |
6 participants
Summary:
Replaced ASSERT_FLOAT_EQ which defaults to fixed kMaxUlps ( = 4-ULP , See gtest-internal.h) with ASSERT_NEAR which lets us set epsilon to 1e-3, (approximately 3 ULPs). This allows for slightly stricter and tunable comparison.
Test Plan:
Before Fix
✗ Fail:
qnnpack:pytorch_qnnpack_testApple - FULLY_CONNECTED_SPARSE_OP_8x1/unit_batch_dynamic_prepacked (0.0s)
'Expected equality of these values:
output_dynamic[i * outputChannels() + c]
Which is: 9.9160004
accumulators_float[i * outputChannels() + c]
Which is: 9.9159956
at 0, 17: reference = 9.9159955978393555, optimized = 9.9160003662109375
After Fix
Everything passes
Rollback Plan:
Differential Revision: D77911682
cc @jgong5 @mingfeima @XiaobingSuper @sanchitintel @ashokei @jingxu10 @jerryzh168