Add new model: Isaac

[zucchini-nlp]

What does this PR do?

Same as #40962 but cleans up the code to match transformers API. Couldn't test due to errors, the integration test is failing atm. Still need to clean the testing file

Ideally this should be working with the following snippet, so users don't need to load image or add imaga tokens themselves

import torch
from transformers import AutoProcessor, IsaacForConditionalGeneration

model_id = "PerceptronAI/Isaac-0.1"
processor = AutoProcessor.from_pretrained(model_id, revision="refs/pr/3",)
model = IsaacForConditionalGeneration.from_pretrained(
    model_id,
    torch_dtype=torch.bfloat16,
    device_map="cuda",
    revision="refs/pr/3",
)


conversation = [
    {

        "role": "user",
        "content": [
            {"type": "text", "text": "Compare the two figures and explain what changed."},
            {"type": "image", "path": "first_image.png"},
            {"type": "image", "path": "second_image.png"},
            ],
    },
]

inputs = processor.apply_chat_template(
    conversation,
    tokenize=True,
    return_dict=True,
    add_generation_prompt=True,
    return_tensors="pt",
).to(model.device)

generated_ids = model.generate(
    **inputs,
    max_new_tokens=256,
    do_sample=False,
)

generated_ids = generated_ids[:, inputs["input_ids"].shape[1] :]
response = processor.batch_decode(generated_ids, skip_special_tokens=True)[0]
print(response)

[github-actions]

[For maintainers] Suggested jobs to run (before merge)

run-slow: auto, isaac

[github-actions]

View the CircleCI Test Summary for this PR:

https://huggingface.co/spaces/transformers-community/circle-ci-viz?pr=45186&sha=a44952