Revise use cases with transformers

[anssiko]

Add new use cases:

• Text-to-image
• Speech Recognition
• Text Generation
• (Image Segmentation now refers to [SegAny] also)

Add new reference models:

• Text-to-image: stable-diffusion-v1-5
• Image segmentation: segment-anything
• Speech-to-text: whisper-tiny.en
• Text generation: t5-small, m2m100_418M, gpt2, llama-2-7b

Remove redundant local reference: [POWERFUL-FEATURES]

---

Preview | Diff

[anssiko]

This is to revise the use cases section with transformers ahead our upcoming "delta" wide review for the new content added since our initial CR in March 2023.

@dontcallmedom please check the use cases are representative of use cases that resonate with our wide review audience. I did not restructure the old material, but plugged in the new transformers-enabled use cases into the existing list. I note it is possible to create a higher-level taxonomy for these use cases but I'll leave that as future work for now. Also check if you can find why PR Preview failed to build. Builds cleanly locally.

@zolkis please check in particular the referenced papers are the canonical ones and the latest, and that the biblio entries are correct.

[bhack]

What about Video transformers?
https://arxiv.org/abs/2201.05991

[zolkis]

Refs (included so far) look good.

[anssiko]

What about Video transformers?
https://arxiv.org/abs/2201.05991

@bhack, it would be helpful if you could open a new issue for the proposal similarly to #524.

[anssiko]

Thanks for your review. I will merge this per discussion on our call https://www.w3.org/2024/01/25-webmachinelearning-minutes.html#t01

These revised use cases help keep the spec in a cohesive state. This is particularly important as we aspire to seek delta wide review outside this group and publish a new CR Snapshot. Please note this is not a complete list of possible use cases. Future updates are expected and contributions are welcome.