Artur Paniukov

Hi @IlyasMoutawwakil, Could you merge the PR? The failed test is network-related:

OpenVINO Tokenizer uses i32 internally too but sets output/input types to i64 for compatibility by default. One can set the` Type.i32` in CLI and `convert_tokenizer` function and we can easily...

Hi @DongChanS, The generate method [works](https://github.com/openvinotoolkit/openvino.genai/blob/3c28e8279ca168ba28a79b50c62ec3b2f61d9f29/src/cpp/include/openvino/genai/llm_pipeline.hpp#L253) with [encoded data](https://github.com/openvinotoolkit/openvino.genai/blob/3c28e8279ca168ba28a79b50c62ec3b2f61d9f29/src/cpp/include/openvino/genai/llm_pipeline.hpp#L27): `ov::Tensor` or [TokenizedSequence](https://github.com/openvinotoolkit/openvino.genai/blob/3c28e8279ca168ba28a79b50c62ec3b2f61d9f29/src/cpp/include/openvino/genai/tokenizer.hpp#L21C1-L24C3). This method will bypass the tokenization step and will return an [encoded result](https://github.com/openvinotoolkit/openvino.genai/blob/3c28e8279ca168ba28a79b50c62ec3b2f61d9f29/src/cpp/include/openvino/genai/llm_pipeline.hpp#L41), not detokenized outputs - so...

> can you also run the code styling Done Delete an import in `pipeline_base.py` to fix this issue: ```shell ruff check . --fix optimum/intel/pipelines/pipeline_base.py:298:54: F811 Redefinition of unused `PreTrainedTokenizerFast` from...

@11happy @Captain-MUDIT There is a [TokenizerPipeline](https://github.com/openvinotoolkit/openvino_tokenizers/blob/df197af63636632f0272c02417aa74fdea8fd23e/python/openvino_tokenizers/tokenizer_pipeline.py#L1291) class for building tokenizer/detokenizer models. The easiest way is to parse the tokenizer data from `.gguf` file, make such a pipeline and get models...

Hi @SearchSavior, The easiest path is to add the `get_prefill_throughput` method to the [PerfMetrics](https://github.com/openvinotoolkit/openvino.genai/blob/5fafe33dbbb7c8ac10ea4b203ef97380f1f329f2/src/cpp/include/openvino/genai/perf_metrics.hpp#L120C31-L120C42) struct. For a single run metric (one generate call), the calculation is straightforward. If you are...