Bf16 Verify Onnx Tests

[richagadgil]

Tests with different ONNX representations than half:

• mean_bf16_test
• instance_norm_bf16 tests

Tests for half that do not seem to have BF16 counterpart:

• mod_test_half

Tests Failing:

• gelu_default_bf16_test

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 92.20%. Comparing base (f9e276b) to head (69c6ebb).

Additional details and impacted files

@@             Coverage Diff             @@
##           develop    #3627      +/-   ##
===========================================
+ Coverage    92.18%   92.20%   +0.01%     
===========================================
  Files          514      514              
  Lines        21780    21783       +3     
===========================================
+ Hits         20078    20084       +6     
+ Misses        1702     1699       -3     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[migraphx-bot]

Test	Batch	Rate new e982b5	Rate old 241e24	Diff	Compare
torchvision-resnet50	64	3,259.45	3,256.24	0.10%	✅
torchvision-resnet50_fp16	64	6,999.78	6,984.56	0.22%	✅
torchvision-densenet121	32	2,432.48	2,434.44	-0.08%	✅
torchvision-densenet121_fp16	32	4,037.37	4,078.22	-1.00%	✅
torchvision-inceptionv3	32	1,629.35	1,629.38	-0.00%	✅
torchvision-inceptionv3_fp16	32	2,746.97	2,742.92	0.15%	✅
cadene-inceptionv4	16	765.04	764.47	0.07%	✅
cadene-resnext64x4	16	810.80	812.01	-0.15%	✅
slim-mobilenet	64	7,383.05	7,468.13	-1.14%	✅
slim-nasnetalarge	64	208.40	208.51	-0.05%	✅
slim-resnet50v2	64	3,441.60	3,440.29	0.04%	✅
bert-mrpc-onnx	8	1,149.29	1,148.20	0.10%	✅
bert-mrpc-tf	1	467.46	464.32	0.68%	✅
pytorch-examples-wlang-gru	1	421.31	426.54	-1.23%	✅
pytorch-examples-wlang-lstm	1	387.37	380.15	1.90%	✅
torchvision-resnet50_1	1	810.54	775.13	4.57%	🔆
cadene-dpn92_1	1	399.38	403.98	-1.14%	✅
cadene-resnext101_1	1	384.77	383.04	0.45%	✅
onnx-taau-downsample	1	345.10	345.85	-0.22%	✅
dlrm-criteoterabyte	1	33.33	33.33	0.02%	✅
dlrm-criteoterabyte_fp16	1	52.76	52.75	0.01%	✅
agentmodel	1	8,476.59	8,145.38	4.07%	🔆
unet_fp16	2	58.88	58.93	-0.09%	✅
resnet50v1_fp16	1	936.77	935.00	0.19%	✅
resnet50v1_int8	1	1,030.91	1,023.68	0.71%	✅
bert_base_cased_fp16	64	1,170.69	1,169.83	0.07%	✅
bert_large_uncased_fp16	32	363.59	363.35	0.07%	✅
bert_large_fp16	1	200.31	195.75	2.33%	✅
distilgpt2_fp16	16	2,198.17	2,196.59	0.07%	✅
yolov5s	1	530.19	534.53	-0.81%	✅
tinyllama	1	43.63	43.39	0.55%	✅
vicuna-fastchat	1	174.13	171.47	1.56%	✅
whisper-tiny-encoder	1	415.70	417.49	-0.43%	✅
whisper-tiny-decoder	1	432.82	427.46	1.25%	✅

Check results before merge 🔆

[migraphx-bot]

     ✅ bert-mrpc-onnx: PASSED: MIGraphX meets tolerance

     ✅ bert-mrpc-tf: PASSED: MIGraphX meets tolerance

     ✅ pytorch-examples-wlang-gru: PASSED: MIGraphX meets tolerance

     ✅ pytorch-examples-wlang-lstm: PASSED: MIGraphX meets tolerance

     ✅ torchvision-resnet50_1: PASSED: MIGraphX meets tolerance

     ✅ cadene-dpn92_1: PASSED: MIGraphX meets tolerance

     ✅ cadene-resnext101_1: PASSED: MIGraphX meets tolerance

     ✅ dlrm-criteoterabyte: PASSED: MIGraphX meets tolerance

     ✅ agentmodel: PASSED: MIGraphX meets tolerance

     ✅ unet: PASSED: MIGraphX meets tolerance

     ✅ resnet50v1: PASSED: MIGraphX meets tolerance

     ✅ bert_base_cased_fp16: PASSED: MIGraphX meets tolerance

🔴bert_large_uncased_fp16: FAILED: MIGraphX is not within tolerance - check verbose output

     ✅ bert_large: PASSED: MIGraphX meets tolerance

     ✅ yolov5s: PASSED: MIGraphX meets tolerance

     ✅ tinyllama: PASSED: MIGraphX meets tolerance

     ✅ vicuna-fastchat: PASSED: MIGraphX meets tolerance

     ✅ whisper-tiny-encoder: PASSED: MIGraphX meets tolerance

     ✅ whisper-tiny-decoder: PASSED: MIGraphX meets tolerance

     ✅ distilgpt2_fp16: PASSED: MIGraphX meets tolerance

[shivadbhavsar]

this looks like it needs to be reverted?