Fix sd3-lite clip-l model swapped outputs

[ikalinic]

The clip-l model from stable-diffusion-3-lite-onnx has swapped outputs, so this results in runtime error because of mismatched shapes. This fix proposes swapping the output tensors for the clip-l outputs if the shape doesn't match the required shape for one of the outputs, making stable-diffusion-3-lite-onnx run successfully.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 92.15%. Comparing base (0860461) to head (542805a).
Report is 17 commits behind head on develop.

Additional details and impacted files

@@             Coverage Diff             @@
##           develop    #3704      +/-   ##
===========================================
- Coverage    92.23%   92.15%   -0.08%     
===========================================
  Files          514      514              
  Lines        21746    21956     +210     
===========================================
+ Hits         20057    20234     +177     
- Misses        1689     1722      +33     

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

[causten]

Just fix the format problem. Ignore the license problem since you didnt touch that code

[migraphx-bot]

Test	Batch	Rate new 542805	Rate old f56b1b	Diff	Compare
torchvision-resnet50	64	3,253.90	3,255.67	-0.05%	✅
torchvision-resnet50_fp16	64	6,987.20	6,983.58	0.05%	✅
torchvision-densenet121	32	2,432.85	2,431.65	0.05%	✅
torchvision-densenet121_fp16	32	4,069.49	4,074.03	-0.11%	✅
torchvision-inceptionv3	32	1,627.16	1,628.91	-0.11%	✅
torchvision-inceptionv3_fp16	32	2,745.23	2,746.14	-0.03%	✅
cadene-inceptionv4	16	765.51	764.54	0.13%	✅
cadene-resnext64x4	16	812.80	813.45	-0.08%	✅
slim-mobilenet	64	7,462.04	7,469.86	-0.10%	✅
slim-nasnetalarge	64	209.00	209.05	-0.03%	✅
slim-resnet50v2	64	3,438.81	3,440.80	-0.06%	✅
bert-mrpc-onnx	8	1,146.07	1,145.22	0.07%	✅
bert-mrpc-tf	1	473.07	476.55	-0.73%	✅
pytorch-examples-wlang-gru	1	419.21	422.09	-0.68%	✅
pytorch-examples-wlang-lstm	1	388.13	394.94	-1.72%	✅
torchvision-resnet50_1	1	814.58	769.40	5.87%	🔆
cadene-dpn92_1	1	407.35	398.97	2.10%	✅
cadene-resnext101_1	1	382.75	383.77	-0.27%	✅
onnx-taau-downsample	1	345.74	345.22	0.15%	✅
dlrm-criteoterabyte	1	33.33	33.32	0.01%	✅
dlrm-criteoterabyte_fp16	1	52.74	52.72	0.04%	✅
agentmodel	1	8,292.15	8,109.71	2.25%	✅
unet_fp16	2	58.91	58.87	0.07%	✅
resnet50v1_fp16	1	926.43	930.54	-0.44%	✅
resnet50v1_int8	1	1,010.40	1,002.61	0.78%	✅
bert_base_cased_fp16	64	1,170.58	1,168.63	0.17%	✅
bert_large_uncased_fp16	32	362.86	363.25	-0.11%	✅
bert_large_fp16	1	200.54	198.22	1.17%	✅
distilgpt2_fp16	16	2,198.60	2,197.80	0.04%	✅
yolov5s	1	524.08	532.80	-1.64%	✅
tinyllama	1	43.35	43.43	-0.17%	✅
vicuna-fastchat	1	174.83	174.20	0.36%	✅
whisper-tiny-encoder	1	417.75	418.04	-0.07%	✅
whisper-tiny-decoder	1	433.22	433.15	0.02%	✅

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