Evgenya Nugmanova

Pre-trained models fail to infer with batch>1 in a standalone code

1

Several Reshape nodes have hard-coded batch dimension value in the 2nd input. Could you please fix the models to preserve batch? This is how all the pre-trained models look like:...

### Details: - *Auto Batch: if plugin is disabled during cmake* ### Tickets: - *88449*

[Good First Issue][TFL FE]: Support GELU operation for TensorFlow Lite

2

### Context OpenVINO component responsible for support of TensorFlow Lite models is called as TensorFlow Lite Frontend (TFL FE). TFL FE converts a model represented in TensorFlow Lite opset to...

Runs symbolic optimizations in MOC

2

### Tickets: - *CVS-136758*

# What does this PR do? OpenVINO exporter sets symbols for model inputs. It helps symbolic propagation and optimizations in OpenVINO. OpenVINO provides symbolic shape inference and optimizations. By default,...

### Details: - *Uses PDPD broadcasting for TF Select v1* ### Tickets: - *N/A*

### Details: - *Introduces operator+ and operator- for ov::Symbol* - *Uses introduced operators in `Add::evaluate_symbol`, `Subtract::evaluate_symbol` and shape_infer of Range operation* Significantly reduces the number of unique symbols in models,...

DO NOT MERGE, FOR VALIDATION ONLY CMP: [Comparing huggingface:main...jane-intel:releases/2024/3 - huggingface/optimum-intel (github.com)](https://github.com/huggingface/optimum-intel/compare/main...jane-intel:releases/2024/3?expand=1) Branch: [jane-intel/optimum-intel at releases/2024/3 (github.com)](https://github.com/jane-intel/optimum-intel/tree/releases/2024/3) main ([819c513](https://github.com/huggingface/optimum-intel/commit/819c51394216eedf9245f91f1a9db897486fff1c)) + [[OV]: load and convert llms in original precision by eaidova...