### Details:
Add tests for the SPDAToPagedAttention transformation to Precommit
1) Check if PagedAttentionExtension node appeared in the model after the
transformation
2) Check 2 last dimensions of 'key_cache.' and 'value_cache.' inputs to
be static.
### Tickets:
- [CVS-138943](https://jira.devtools.intel.com/browse/CVS-138943)
Signed-off-by: Andrii Staikov <andrii.staikov@intel.com>
### Details:
- *After support for i64 was added in frontend `aten::adaptive_avg_pool`
started to fail if shape input is list.*
### Tickets:
- *CVS-141335*
**Details:** LookupTableImport operation node can be terminating in a
graph and needs to be preserved in a graph for further HashTable
operation resolving.
**Ticket:** TBD
---------
Signed-off-by: Kazantsev, Roman <roman.kazantsev@intel.com>
**Details:** Fix model input and output names in case absence of serving
signature. mil-nce model from TF Hub does not have default signature
("serving") so use user names from other signatures to resolve internal
tensor names.
Also, it adds validation for `read_model` scenario.
**Tickets:** 139245, 124401
---------
Signed-off-by: Kazantsev, Roman <roman.kazantsev@intel.com>
### Details:
- *Added `kornia` dependency for lightglue model*
- *Disable GFPGAN since it doesn't support latest `torchvision`*
- *Remove `transfo-xl-wt103` as not supported on latest `transformers`*
- *Remove XFAIL from `facebook/levit-128S`*
### Tickets:
- *ticket-id*
### Details:
- *Fix translation for `aten.copy.default`*
- *Add support for `aten.rand.default`*
- *Support `vit-mae` model*
- *Update hf model list with newly added models since list creation*
### Tickets:
- *CVS-133732*
### Details:
There were two problems with memory sharing, which affected HF models.
`value()` method of class `ResourceVariable` copies value of variable in
some cases. This results in sharing of memory with a copied value, which
is deallocated when `GraphIterator` finishes its work.
Replaced `value()` with `read_value_no_copy()` which should not copy
value in any case.
Another problem is that original model inference moves weights to new
location which leads to corruption of shared memory in OV model.
So the original model should be inferred after the OV inference to
prevent any changes in original model.
### Tickets:
- 134787, 134408
---------
Co-authored-by: Roman Kazantsev <roman.kazantsev@intel.com>
### Details:
- Tests for Hugging Face models fail when output structure has
dictionary of tuples. For such case comparison method was changed to
compare outputs using original model output signature, which has inner
tensor names that can be matched with OV output.
### Tickets:
- 134405
---------
Co-authored-by: Roman Kazantsev <roman.kazantsev@intel.com>
New PR due to merge
conflicts(https://github.com/openvinotoolkit/openvino/pull/22917)
Updated the copyright year from 2018-2023 to 2018-2024 in all openvino
files
Ref. to script: CVS-101144
Command used:
```bash
git grep -lz '2018-2023 Intel Corporation' | xargs -0 sed -i '' -e 's/2018-2023 Intel Corporation/2018-2024 Intel Corporation/g'
```
### Details:
- *Update list of timm models to comply with new version of timm*
- *Run timm models and torchvision models in trace and export model in
parallel*
### Tickets:
- *ticket-id*
### Details:
- *Add support for `aten::inverse`*
- *Tests: couldn't replicate the operation in layer tests, but it exists
in TPSMM model, so test for this model is included*
### Tickets:
- *CVS-123613*
**Details:** Fix bad cast for tf.TensorShape to ov.PartialShape. The
error occurred in mobile-object-localizer-v1 model.
**Ticket:** CVS-132455
---------
Signed-off-by: Kazantsev, Roman <roman.kazantsev@intel.com>
* [PT FE] torch.export support
* Apply code style
* Fix build
* Support torch <2.2
* Support more operations
* Update tests/model_hub_tests/torch_tests/torch_utils.py
* Support for model operations
* Support is_causal as kwarg for SDPA
* Update src/frontends/pytorch/src/op/addcmul.cpp
* Update tests/model_hub_tests/torch_tests/test_timm.py
* Support only decoder passed to convert_model
* Fix tests
* Update src/bindings/python/src/openvino/frontend/pytorch/fx_decoder.py
* Update src/bindings/python/src/openvino/frontend/pytorch/fx_decoder.py
Co-authored-by: Ekaterina Aidova <ekaterina.aidova@intel.com>
* Apply suggestions from code review
* Apply suggestions from code review
* Improve testing with caching model
* Apply suggestions from code review
---------
Co-authored-by: Ekaterina Aidova <ekaterina.aidova@intel.com>
* fixed list index out of range error in node_decoder.py
* Improved and Modified as per review feedback
- Added a Model in the precommit_models for precommit validation.
- Removed skip and xfail marrks for the issue (129232 IndexError list
index out of range for Squeeze)
- Added explicilty return None in node_decoder.py
* Update tests/model_hub_tests/tf_hub_tests/nightly_models
* Update tests/model_hub_tests/tf_hub_tests/nightly_models
* Update tests/model_hub_tests/tf_hub_tests/nightly_models
* Apply suggestions from code review
* Update tests/model_hub_tests/tf_hub_tests/precommit_models
* Update tests/model_hub_tests/tf_hub_tests/precommit_models
---------
Co-authored-by: Roman Kazantsev <roman.kazantsev@intel.com>
* [TF Hub] Mark up models failing due to unsupported operations
Signed-off-by: Kazantsev, Roman <roman.kazantsev@intel.com>
* Update tests/model_hub_tests/tf_hub_tests/nightly_models
---------
Signed-off-by: Kazantsev, Roman <roman.kazantsev@intel.com>
* [TF FE] Add testing model with tokenizer to precommit
Signed-off-by: Kazantsev, Roman <roman.kazantsev@intel.com>
* Fix typo
---------
Signed-off-by: Kazantsev, Roman <roman.kazantsev@intel.com>
* [TF Hub] Mark up models failing with IndexError for Squeeze
Signed-off-by: Kazantsev, Roman <roman.kazantsev@intel.com>
* Add xfail for nightly
Signed-off-by: Kazantsev, Roman <roman.kazantsev@intel.com>
---------
Signed-off-by: Kazantsev, Roman <roman.kazantsev@intel.com>
* [TF Hub] Mark tests failing with InvalidArgumentError
Signed-off-by: Kazantsev, Roman <roman.kazantsev@intel.com>
* [TF Hub] Mark test models failing with AttributeError
Signed-off-by: Kazantsev, Roman <roman.kazantsev@intel.com>
---------
Signed-off-by: Kazantsev, Roman <roman.kazantsev@intel.com>
Maxim Vafin