Change driver verify to check for fp16 and --fp16

[CharlieL7]

Change the tolerances based on what we find.
Use std::optional.

[codecov]

Codecov Report

Merging #2334 (abdb00e) into develop (94bda24) will not change coverage.
The diff coverage is n/a.

❗ Current head abdb00e differs from pull request most recent head 5a3ba68. Consider uploading reports for the commit 5a3ba68 to get more accurate results

@@           Coverage Diff            @@
##           develop    #2334   +/-   ##
========================================
  Coverage    91.35%   91.35%           
========================================
  Files          438      438           
  Lines        16465    16465           
========================================
  Hits         15041    15041           
  Misses        1424     1424           

[migraphx-bot]

Test	Batch	Rate new acd4c5	Rate old 650ba4	Diff	Compare
torchvision-resnet50	64	2,853.78	2,855.07	-0.05%	✅
torchvision-resnet50_fp16	64	6,481.63	6,485.89	-0.07%	✅
torchvision-densenet121	32	2,105.15	2,099.19	0.28%	✅
torchvision-densenet121_fp16	32	3,693.70	3,685.09	0.23%	✅
torchvision-inceptionv3	32	1,597.46	1,598.81	-0.08%	✅
torchvision-inceptionv3_fp16	32	2,594.94	2,594.10	0.03%	✅
cadene-inceptionv4	16	707.61	708.32	-0.10%	✅
cadene-resnext64x4	16	698.38	698.74	-0.05%	✅
slim-mobilenet	64	8,369.15	8,357.32	0.14%	✅
slim-nasnetalarge	64	227.10	227.33	-0.10%	✅
slim-resnet50v2	64	2,679.07	2,677.94	0.04%	✅
bert-mrpc-onnx	8	825.07	825.54	-0.06%	✅
bert-mrpc-tf	1	388.27	387.74	0.14%	✅
pytorch-examples-wlang-gru	1	288.15	296.73	-2.89%	✅
pytorch-examples-wlang-lstm	1	303.44	307.96	-1.47%	✅
torchvision-resnet50_1	1	603.81	605.56	-0.29%	✅
torchvision-inceptionv3_1	1	339.31	337.94	0.40%	✅
cadene-dpn92_1	1	398.60	396.52	0.53%	✅
cadene-resnext101_1	1	329.34	329.16	0.05%	✅
slim-vgg16_1	1	464.09	464.46	-0.08%	✅
slim-mobilenet_1	1	2,080.10	2,075.27	0.23%	✅
slim-inceptionv4_1	1	217.58	217.23	0.16%	✅
onnx-taau-downsample	1	306.26	306.31	-0.02%	✅
dlrm-criteoterabyte	1	21.70	21.72	-0.07%	✅
dlrm-criteoterabyte_fp16	1	40.76	40.75	0.03%	✅
agentmodel	1	5,712.95	5,825.40	-1.93%	✅
unet_fp16	2	55.99	56.00	-0.03%	✅
resnet50v1_fp16	1	935.89	935.59	0.03%	✅
bert_base_cased_fp16	64	971.24	971.23	0.00%	✅
bert_large_uncased_fp16	32	305.20	305.23	-0.01%	✅
bert_large_fp16	1	166.75	167.09	-0.21%	✅
distilgpt2_fp16	16	1,278.86	1,278.55	0.02%	✅

This build is OK for merge ✅

[migraphx-bot]

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

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

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

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

    :white_check_mark:torchvision-resnet50_1: PASSED: MIGraphX meets tolerance

🔴torchvision-inceptionv3_1: FAILED: MIGraphX is not within tolerance - check verbose output

    :white_check_mark:cadene-dpn92_1: PASSED: MIGraphX meets tolerance

    :white_check_mark:cadene-resnext101_1: PASSED: MIGraphX meets tolerance

    :white_check_mark:slim-vgg16_1: PASSED: MIGraphX meets tolerance

    :white_check_mark:slim-mobilenet_1: PASSED: MIGraphX meets tolerance

🔴slim-inceptionv4_1: FAILED: MIGraphX is not within tolerance - check verbose output

    :white_check_mark:dlrm-criteoterabyte: PASSED: MIGraphX meets tolerance

    :white_check_mark:agentmodel: PASSED: MIGraphX meets tolerance

    :white_check_mark:unet: PASSED: MIGraphX meets tolerance

    :white_check_mark:resnet50v1: PASSED: MIGraphX meets tolerance

🔴bert_base_cased_fp16: FAILED: MIGraphX is not within tolerance - check verbose output

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

    :white_check_mark:bert_large: PASSED: MIGraphX meets tolerance

🔴distilgpt2_fp16: FAILED: MIGraphX is not within tolerance - check verbose output

[causten]

@CharlieL7 can you check if the tolerances are valid, I see two models failing that were not in the past

[CharlieL7]

@CharlieL7 can you check if the tolerances are valid, I see two models failing that were not in the past

@causten Those tests are using the accuracy_checker script, should be unrelated to the changes in this PR. Issue originates from #2274.

[pfultz2]

Actually this should return std::optional<T>: static std::optional<T> apply(const std::string& x) { return value_parser<T>::apply(x); }.