version_rec.text

DinoV2:

Should use train.py and this new versions of transformers support llama's flash attention.

dependencies = [
    "torch==2.0.1", "torchvision==0.15.2",
    "transformers==4.38.2", "tokenizers==0.15.2", "sentencepiece==0.1.99", "shortuuid",
    "accelerate==0.27.2", "peft==0.4.0", "bitsandbytes==0.41.0",
    "pydantic<2,>=1", "markdown2[all]", "numpy", "scikit-learn==1.2.2",
    "gradio==3.35.2", "gradio_client==0.2.9",
    "requests", "httpx==0.24.0", "uvicorn", "fastapi",
    "einops==0.6.1", "einops-exts==0.0.4", "timm==0.6.13",
    "diffusers==0.19.0", "xformers==0.0.22", "invisible-watermark==0.2.0"
]

[project.optional-dependencies]
train = ["deepspeed==0.12.6", "ninja"]


OpenCLIP:

dependencies = [
    "torch==2.0.1", "torchvision==0.15.2",
    "transformers==4.31.0", "tokenizers>=0.12.1,<0.14", "sentencepiece==0.1.99", "shortuuid",
    "accelerate==0.21.0", "peft==0.4.0", "bitsandbytes==0.41.0",
    "pydantic<2,>=1", "markdown2[all]", "numpy", "scikit-learn==1.2.2",
    "gradio==3.35.2", "gradio_client==0.2.9",
    "requests", "httpx==0.24.0", "uvicorn", "fastapi",
    "einops==0.6.1", "einops-exts==0.0.4", "timm==0.6.13",
    "diffusers==0.19.0", "xformers==0.0.22", "invisible-watermark==0.2.0"
]

[project.optional-dependencies]
train = ["deepspeed==0.9.5", "ninja"]