Docs: Resolve various typos

[tomaarsen]

Hello!

Pull Request overview

• Fix various typos.

Details

I found the first few manually in the README, and then found the rest via codespell.

Note

There's another typo which seems more critical to the behaviour of the package, so I did not change that one:

optimum-benchmark/optimum_benchmark/benchmarks/training/benchmark.py

Line 53 in 90ad2ea

"overall_training.throughput(samles/s)": (trainer_state.overall_training_samples_per_second),

samles should be samples. Because of this, a bunch of the csv's also have this samles: https://github.com/huggingface/optimum-benchmark/blob/90ad2ea78d23ad59ef23ee9552a1c484706ec336/examples/training-llamas/experiments/A100-80GB/llama_peft%2Bbnb_batch_size(2)/1/training_results.csv

I'll leave this one for you to fix, as I'm not sure how it all connects together.

• Tom Aarsen

[IlyasMoutawwakil]

Thanks a lott! definitely need to integrate a codespell workflow!

[IlyasMoutawwakil]

@tomaarsen can you run make style

[tomaarsen]

Of course! Done.

And don't bother with integrating codespell in a CI or something - it makes a lot of mistakes too, so it requires a manual look :)

[tomaarsen]

Oh, I'm probably using a different version of black

[IlyasMoutawwakil]

Thanks, will correct the remaining one later