Benchmarks: Add benchmark: Megatron-LM/Megatron-Deepspeed GPT pretrain benchmark

[yukirora]

Description
Megatron-LM/Megatron-Deepspeed GPT pretrain benchmark

[codecov]

Codecov Report

Attention: 65 lines in your changes are missing coverage. Please review.

Comparison is base (254ea7f) 86.36% compared to head (5a47989) 85.98%.

Files	Patch %	Lines
...bench/benchmarks/model_benchmarks/megatron_gpt3.py	76.89%	61 Missing ⚠️
...perbench/benchmarks/model_benchmarks/model_base.py	55.55%	4 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #582      +/-   ##
==========================================
- Coverage   86.36%   85.98%   -0.38%     
==========================================
  Files          96       97       +1     
  Lines        6573     6845     +272     
==========================================
+ Hits         5677     5886     +209     
- Misses        896      959      +63     

Flag	Coverage Δ
cpu-python3.6-unit-test	71.99% <76.36%> (+0.20%)	⬆️
cpu-python3.7-unit-test	71.99% <76.36%> (+0.20%)	⬆️
cpu-python3.8-unit-test	72.40% <76.36%> (+0.18%)	⬆️
cuda-unit-test	84.00% <76.36%> (-0.30%)	⬇️
directx-unit-test	35.31% <31.63%> (-0.17%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

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