[7/7] Multimodal datasets (The Cauldron, LLaVA-Instruct-150K)

[RdoubleA]

Changelog

• Add two example multimodal dataset builders, each with their own dataset transform: The Cauldron and LLaVA-Instruct-150K. Both require slightly different preprocessing and serve as good examples
• Add a utility to quickly map raw text with image tags into what's expected for Message content field
• Upgrade ShareGPTToMessages with support for an image column

Test plan

• Unit test for The Cauldron, LLaVA Instruct datasets
• Unit test for split_text_by_image_tag
• TODO: update unit test for ShareGPTToMessages

Docs

TODO: update docstrings and API ref

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/torchtune/1158

• 📄 Preview Python docs built from this PR

Note: Links to docs will display an error until the docs builds have been completed.

❌ 3 New Failures, 5 Cancelled Jobs

As of commit 8530958 with merge base 9e65fa9 ():

NEW FAILURES - The following jobs have failed:

• GPU tests / gpu_test (3.10, stable) (gh)
  tests/torchtune/datasets/test_the_cauldron_dataset.py::TestTheCauldronDataset::test_label_masking
• GPU tests / gpu_test (3.11, stable) (gh)
  tests/torchtune/datasets/test_the_cauldron_dataset.py::TestTheCauldronDataset::test_label_masking
• Unit Test / unit_tests (3.11) (gh)
  tests/torchtune/datasets/test_the_cauldron_dataset.py::TestTheCauldronDataset::test_label_masking

CANCELLED JOBS - The following jobs were cancelled. Please retry:

• GPU tests / gpu_test (3.8, stable) (gh)
• GPU tests / gpu_test (3.9, stable) (gh)
  tests/torchtune/datasets/test_the_cauldron_dataset.py::TestTheCauldronDataset::test_label_masking
• Unit Test / unit_tests (3.10) (gh)
• Unit Test / unit_tests (3.8) (gh)
• Unit Test / unit_tests (3.9) (gh)
  ##[error]The operation was canceled.

This comment was automatically generated by Dr. CI and updates every 15 minutes.