add int4 and int8 checks

[DaniAffCH]

Changes

This PR addresses #2565

• Added a function to retrieve the number of int4 and int8 ops within the Weight Compression tests.
• Added check to ensure that the number of int4 and int8 ops match the reference values
• Added reference values for int4 and int8 ops

Related tickets

Ticket: 132562

Tests

Tested running:

pytest test_quantize_conformance.py::test_weight_compression[tinyllama_data_aware_awq_stateful_backend_OV] --data=

[DaniAffCH]

Unfortunately, my laptop memory was not enough to execute all the models to obtain the reference values. I could run only tinyllama_data_free_backend_OV as it is slightly lighter than the others.

I would appreciate it if you could run the remaining models, pointing out the reference values for num_int4 and num_int8 so I can add them to the reference.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 77.93%. Comparing base (b134935) to head (785aad2).

Additional details and impacted files

@@             Coverage Diff              @@
##           develop    #2617       +/-   ##
============================================
- Coverage    91.16%   77.93%   -13.23%     
============================================
  Files          494      494               
  Lines        45401    45401               
============================================
- Hits         41388    35382     -6006     
- Misses        4013    10019     +6006     

see 105 files with indirect coverage changes

Flag	Coverage Δ
COMMON	?
ONNX	?
OPENVINO	?
TENSORFLOW	30.10% <ø> (ø)
TORCH	65.95% <ø> (ø)

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

Components	Coverage Δ
common	88.28% <ø> (-5.47%)	⬇️
torch	93.49% <ø> (-0.01%)	⬇️
tensorflow	93.74% <ø> (ø)
onnx	0.00% <ø> (-93.09%)	⬇️
openvino	25.70% <ø> (-68.47%)	⬇️
ptq	53.06% <ø> (-37.03%)	⬇️

[ljaljushkin]

I noticed some bug
references numbers are different for data-free, but test is passing and results file has different numbers

when run separately, results file has corrent numbers

[DaniAffCH]

Nice catch @ljaljushkin! I think the issue is fixed now

[andrey-churkin]

@DaniAffCH Could you please rebase your branch to run CI one more time? Thanks.

[DaniAffCH]

Done, all the CI tests passed.

[ljaljushkin]

44 build

[ljaljushkin]

Thanks, @DaniAffCH! 🎉