Image Classification Guide Missing Dependency and Addition for Local Models

[Tanman2001]

System Info

• transformers version: 4.39.2
• Platform: Windows-10-10.0.22631-SP0
• Python version: 3.11.1
• Huggingface_hub version: 0.22.1
• Safetensors version: 0.4.2
• Accelerate version: 0.28.0
• Accelerate config: not found
• PyTorch version (GPU?): 2.2.2+cu118 (True)
• Tensorflow version (GPU?): not installed (NA)
• Flax version (CPU?/GPU?/TPU?): not installed (NA)
• Jax version: not installed
• JaxLib version: not installed
• Using GPU in script?:
• Using distributed or parallel set-up in script?:

Who can help?

No response

Information

• The official example scripts
• My own modified scripts

Tasks

• An officially supported task in the examples folder (such as GLUE/SQuAD, ...)
• My own task or dataset (give details below)

Reproduction

Tested using PyTorch only.

• Missing pip install scikit-learn

1. Follow Image Classification guide here: https://huggingface.co/docs/transformers/tasks/image_classification
2. Warning on accuracy = evaluate.load("accuracy") saying scikit-learn is missing.
3. Later training crashes due to missing module.

Additional recommendation:

Users should be able to use their model for inference even when they are unable to or it is not preferable for the user to push their model to the hub.

Add to guide under inference, in the pipelining section:
classifier = pipeline("image-classification", model="<path of checkpoint folder>")
as an option to use a locally stored checkpoint of their model for inference.

Expected behavior

Following guide as written should not result in crash.

Additionally, users should know how to use the model when it is saved locally (as checkpointing automatically does).

[ArthurZucker]

Very related to #30058