Lyalyushkin Nikolay

### Changes Introduced --torch-with-cuda11 option for 3rd party sanity tests to install torch compatible with GTX3090. ### Reason for changes There's issue with torch installation inside a 3rd party test....

Allow to setup compression modules for the training-only modules

1

It's needed for elastic width in case of Aux branches, e.g. in InceptionV3. We still don't need to compress these branches, but we need Mixin modules with pre_ops to propagate...

> Also, as I've discovered just now, `tensorflow.python.ops.gen_array_ops.fake_quant_with_min_max_vars_per_channel` uses `num_bits` as a corresponding keyword argument. Wouldn't it be better to be aligned with this as well? ok, I will rename...

Dummy change to test CI

1

### Changes dummy changes ### Reason for changes to test stopping of outdated builds ### Related tickets 82400 ### Tests n/a

[Good First Issue][NNCF]: dump the ignored scope more gracefully

4

### Context NNCF saves parameters of the compression to [OpenVINO IR](https://docs.openvino.ai/2024/documentation/openvino-ir-format.html). It's done in [dump_parameters](https://github.com/openvinotoolkit/nncf/blob/edd397b01f8c45b976785fa67d425467d80835ae/nncf/openvino/rt_info.py#L20) method using [set_rt_info](https://docs.openvino.ai/2024/api/ie_python_api/_autosummary/openvino.runtime.Model.html#openvino.runtime.Model.set_rt_info) API. But the problem is that ignored scope is saved not gracefully...

Use latest optimum to not call deprecated shared_memory API

4

### Changes use the latest release of optimum-intel==1.15.2 ### Reason for changes The previously used version of optimum had a call of deprecated API that was removed in 2024 release...

[Good First Issue][NNCF]: check number of u8, u4 constants in weight compression tests

4

### Context The issue proposes adding additional check to [post-training compression conformance suite](https://github.com/openvinotoolkit/nncf/blob/develop/tests/post_training/README.md), particularly for [weight compression part](blob/develop/tests/post_training/test_quantize_conformance.py#L248). Currently, it counts number of `FakeQuantize` ops only: [code](https://github.com/openvinotoolkit/nncf/blob/develop/tests/post_training/pipelines/base.py#L381) But it's only...

[Good First Issue][NNCF]: fix invalid error reporting in JSON schema

19

### Context Configuration for NNCF algorithms can be defined in JSON format. For instance, [config.json].(https://github.com/openvinotoolkit/nncf/blob/develop/examples/torch/classification/configs/sparsity_quantization/inception_v3_imagenet_rb_sparsity_int8.json#L17-L38) Compression section in these configuration files are validated using [jsonschema.validate](https://github.com/openvinotoolkit/nncf/blob/develop/nncf/config/schema.py#L163C9-L163C28) But there's some bug with...

### Changes Explicit enabling of dynamic quantization of activations and updating metrics in the weight compression examples. It's possible to adapt compression examples for new openvino-nightly by explicitly turning off...

Fix GPTQ for inputs with batch size != 1 and with seq len == 1

1

### Changes GPTQ correctly processes inputs with batch size != 1 and with batch size and sequence length equal 1. Also changed the errors we are raising in NNCF from...