diff --git a/pyproject.toml b/pyproject.toml index 49347c8ab0..d14a392b8d 100644 --- a/pyproject.toml +++ b/pyproject.toml @@ -32,11 +32,17 @@ include-package-data = false exclude = ["images*"] [project.optional-dependencies] +triton = [ + "triton @ https://github.com/woct0rdho/triton-windows/releases/download/v3.1.0-windows.post5/triton-3.1.0-cp39-cp39-win_amd64.whl ; python_version=='3.9' and platform_system == 'Windows'", + "triton @ https://github.com/woct0rdho/triton-windows/releases/download/v3.1.0-windows.post5/triton-3.1.0-cp310-cp310-win_amd64.whl ; python_version=='3.10' and platform_system == 'Windows'", + "triton @ https://github.com/woct0rdho/triton-windows/releases/download/v3.1.0-windows.post5/triton-3.1.0-cp311-cp311-win_amd64.whl ; python_version=='3.11' and platform_system == 'Windows'", + "triton @ https://github.com/woct0rdho/triton-windows/releases/download/v3.1.0-windows.post5/triton-3.1.0-cp312-cp312-win_amd64.whl ; python_version=='3.12' and platform_system == 'Windows'", +] huggingface = [ - "unsloth_zoo>=2024.11.8", + "unsloth_zoo>=2024.12.4", "packaging", "tyro", - "transformers>=4.46.1,<=4.46.3", + "transformers>=4.46.1,!=4.47.0", "datasets>=2.16.0", "sentencepiece>=0.2.0", "tqdm", @@ -49,88 +55,113 @@ huggingface = [ "protobuf<4.0.0", "huggingface_hub", "hf_transfer", + "unsloth[triton]", ] cu118only = [ - "xformers @ https://download.pytorch.org/whl/cu118/xformers-0.0.22.post7%2Bcu118-cp39-cp39-manylinux2014_x86_64.whl ; python_version=='3.9'", - "xformers @ https://download.pytorch.org/whl/cu118/xformers-0.0.22.post7%2Bcu118-cp310-cp310-manylinux2014_x86_64.whl ; python_version=='3.10'", - "xformers @ https://download.pytorch.org/whl/cu118/xformers-0.0.22.post7%2Bcu118-cp311-cp311-manylinux2014_x86_64.whl ; python_version=='3.11'", + "xformers @ https://download.pytorch.org/whl/cu118/xformers-0.0.22.post7%2Bcu118-cp39-cp39-manylinux2014_x86_64.whl ; python_version=='3.9' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu118/xformers-0.0.22.post7%2Bcu118-cp310-cp310-manylinux2014_x86_64.whl ; python_version=='3.10' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu118/xformers-0.0.22.post7%2Bcu118-cp311-cp311-manylinux2014_x86_64.whl ; python_version=='3.11' and platform_system == 'Linux'", ] cu121only = [ - "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.22.post7-cp39-cp39-manylinux2014_x86_64.whl ; python_version=='3.9'", - "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.22.post7-cp310-cp310-manylinux2014_x86_64.whl ; python_version=='3.10'", - "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.22.post7-cp311-cp311-manylinux2014_x86_64.whl ; python_version=='3.11'", + "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.22.post7-cp39-cp39-manylinux2014_x86_64.whl ; python_version=='3.9' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.22.post7-cp310-cp310-manylinux2014_x86_64.whl ; python_version=='3.10' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.22.post7-cp311-cp311-manylinux2014_x86_64.whl ; python_version=='3.11' and platform_system == 'Linux'", ] cu118onlytorch211 = [ - "xformers @ https://download.pytorch.org/whl/cu118/xformers-0.0.23%2Bcu118-cp39-cp39-manylinux2014_x86_64.whl ; python_version=='3.9'", - "xformers @ https://download.pytorch.org/whl/cu118/xformers-0.0.23%2Bcu118-cp310-cp310-manylinux2014_x86_64.whl ; python_version=='3.10'", - "xformers @ https://download.pytorch.org/whl/cu118/xformers-0.0.23%2Bcu118-cp311-cp311-manylinux2014_x86_64.whl ; python_version=='3.11'", + "xformers @ https://download.pytorch.org/whl/cu118/xformers-0.0.23%2Bcu118-cp39-cp39-manylinux2014_x86_64.whl ; python_version=='3.9' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu118/xformers-0.0.23%2Bcu118-cp310-cp310-manylinux2014_x86_64.whl ; python_version=='3.10' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu118/xformers-0.0.23%2Bcu118-cp311-cp311-manylinux2014_x86_64.whl ; python_version=='3.11' and platform_system == 'Linux'", ] cu121onlytorch211 = [ - "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.23-cp39-cp39-manylinux2014_x86_64.whl ; python_version=='3.9'", - "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.23-cp310-cp310-manylinux2014_x86_64.whl ; python_version=='3.10'", - "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.23-cp311-cp311-manylinux2014_x86_64.whl ; python_version=='3.11'", + "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.23-cp39-cp39-manylinux2014_x86_64.whl ; python_version=='3.9' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.23-cp310-cp310-manylinux2014_x86_64.whl ; python_version=='3.10' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.23-cp311-cp311-manylinux2014_x86_64.whl ; python_version=='3.11' and platform_system == 'Linux'", ] cu118onlytorch212 = [ - "xformers @ https://download.pytorch.org/whl/cu118/xformers-0.0.23.post1%2Bcu118-cp39-cp39-manylinux2014_x86_64.whl ; python_version=='3.9'", - "xformers @ https://download.pytorch.org/whl/cu118/xformers-0.0.23.post1%2Bcu118-cp310-cp310-manylinux2014_x86_64.whl ; python_version=='3.10'", - "xformers @ https://download.pytorch.org/whl/cu118/xformers-0.0.23.post1%2Bcu118-cp311-cp311-manylinux2014_x86_64.whl ; python_version=='3.11'", + "xformers @ https://download.pytorch.org/whl/cu118/xformers-0.0.23.post1%2Bcu118-cp39-cp39-manylinux2014_x86_64.whl ; python_version=='3.9' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu118/xformers-0.0.23.post1%2Bcu118-cp310-cp310-manylinux2014_x86_64.whl ; python_version=='3.10' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu118/xformers-0.0.23.post1%2Bcu118-cp311-cp311-manylinux2014_x86_64.whl ; python_version=='3.11' and platform_system == 'Linux'", ] cu121onlytorch212 = [ - "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.23.post1-cp39-cp39-manylinux2014_x86_64.whl ; python_version=='3.9'", - "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.23.post1-cp310-cp310-manylinux2014_x86_64.whl ; python_version=='3.10'", - "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.23.post1-cp311-cp311-manylinux2014_x86_64.whl ; python_version=='3.11'", + "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.23.post1-cp39-cp39-manylinux2014_x86_64.whl ; python_version=='3.9' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.23.post1-cp310-cp310-manylinux2014_x86_64.whl ; python_version=='3.10' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.23.post1-cp311-cp311-manylinux2014_x86_64.whl ; python_version=='3.11' and platform_system == 'Linux'", ] cu118onlytorch220 = [ - "xformers @ https://download.pytorch.org/whl/cu118/xformers-0.0.24%2Bcu118-cp39-cp39-manylinux2014_x86_64.whl ; python_version=='3.9'", - "xformers @ https://download.pytorch.org/whl/cu118/xformers-0.0.24%2Bcu118-cp310-cp310-manylinux2014_x86_64.whl ; python_version=='3.10'", - "xformers @ https://download.pytorch.org/whl/cu118/xformers-0.0.24%2Bcu118-cp311-cp311-manylinux2014_x86_64.whl ; python_version=='3.11'", + "xformers @ https://download.pytorch.org/whl/cu118/xformers-0.0.24%2Bcu118-cp39-cp39-manylinux2014_x86_64.whl ; python_version=='3.9' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu118/xformers-0.0.24%2Bcu118-cp310-cp310-manylinux2014_x86_64.whl ; python_version=='3.10' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu118/xformers-0.0.24%2Bcu118-cp311-cp311-manylinux2014_x86_64.whl ; python_version=='3.11' and platform_system == 'Linux'", ] cu121onlytorch220 = [ - "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.24-cp39-cp39-manylinux2014_x86_64.whl ; python_version=='3.9'", - "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.24-cp310-cp310-manylinux2014_x86_64.whl ; python_version=='3.10'", - "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.24-cp311-cp311-manylinux2014_x86_64.whl ; python_version=='3.11'", + "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.24-cp39-cp39-manylinux2014_x86_64.whl ; python_version=='3.9' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.24-cp310-cp310-manylinux2014_x86_64.whl ; python_version=='3.10' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.24-cp311-cp311-manylinux2014_x86_64.whl ; python_version=='3.11' and platform_system == 'Linux'", ] cu118onlytorch230 = [ - "xformers @ https://download.pytorch.org/whl/cu118/xformers-0.0.27%2Bcu118-cp39-cp39-manylinux2014_x86_64.whl ; python_version=='3.9'", - "xformers @ https://download.pytorch.org/whl/cu118/xformers-0.0.27%2Bcu118-cp310-cp310-manylinux2014_x86_64.whl ; python_version=='3.10'", - "xformers @ https://download.pytorch.org/whl/cu118/xformers-0.0.27%2Bcu118-cp311-cp311-manylinux2014_x86_64.whl ; python_version=='3.11'", - "xformers @ https://download.pytorch.org/whl/cu118/xformers-0.0.27%2Bcu118-cp312-cp312-manylinux2014_x86_64.whl ; python_version=='3.12'", + "xformers @ https://download.pytorch.org/whl/cu118/xformers-0.0.27%2Bcu118-cp39-cp39-manylinux2014_x86_64.whl ; python_version=='3.9' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu118/xformers-0.0.27%2Bcu118-cp310-cp310-manylinux2014_x86_64.whl ; python_version=='3.10' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu118/xformers-0.0.27%2Bcu118-cp311-cp311-manylinux2014_x86_64.whl ; python_version=='3.11' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu118/xformers-0.0.27%2Bcu118-cp312-cp312-manylinux2014_x86_64.whl ; python_version=='3.12' and platform_system == 'Linux'", ] cu121onlytorch230 = [ - "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.27-cp39-cp39-manylinux2014_x86_64.whl ; python_version=='3.9'", - "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.27-cp310-cp310-manylinux2014_x86_64.whl ; python_version=='3.10'", - "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.27-cp311-cp311-manylinux2014_x86_64.whl ; python_version=='3.11'", - "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.27-cp312-cp312-manylinux2014_x86_64.whl ; python_version=='3.12'", + "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.27-cp39-cp39-manylinux2014_x86_64.whl ; python_version=='3.9' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.27-cp310-cp310-manylinux2014_x86_64.whl ; python_version=='3.10' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.27-cp311-cp311-manylinux2014_x86_64.whl ; python_version=='3.11' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.27-cp312-cp312-manylinux2014_x86_64.whl ; python_version=='3.12' and platform_system == 'Linux'", ] cu118onlytorch240 = [ - "xformers @ https://download.pytorch.org/whl/cu118/xformers-0.0.27.post2%2Bcu118-cp39-cp39-manylinux2014_x86_64.whl ; python_version=='3.9'", - "xformers @ https://download.pytorch.org/whl/cu118/xformers-0.0.27.post2%2Bcu118-cp310-cp310-manylinux2014_x86_64.whl ; python_version=='3.10'", - "xformers @ https://download.pytorch.org/whl/cu118/xformers-0.0.27.post2%2Bcu118-cp311-cp311-manylinux2014_x86_64.whl ; python_version=='3.11'", - "xformers @ https://download.pytorch.org/whl/cu118/xformers-0.0.27.post2%2Bcu118-cp312-cp312-manylinux2014_x86_64.whl ; python_version=='3.12'", + "xformers @ https://download.pytorch.org/whl/cu118/xformers-0.0.27.post2%2Bcu118-cp39-cp39-manylinux2014_x86_64.whl ; python_version=='3.9' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu118/xformers-0.0.27.post2%2Bcu118-cp310-cp310-manylinux2014_x86_64.whl ; python_version=='3.10' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu118/xformers-0.0.27.post2%2Bcu118-cp311-cp311-manylinux2014_x86_64.whl ; python_version=='3.11' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu118/xformers-0.0.27.post2%2Bcu118-cp312-cp312-manylinux2014_x86_64.whl ; python_version=='3.12' and platform_system == 'Linux'", ] cu121onlytorch240 = [ - "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.28.post1-cp39-cp39-manylinux_2_28_x86_64.whl ; python_version=='3.9'", - "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.28.post1-cp310-cp310-manylinux_2_28_x86_64.whl ; python_version=='3.10'", - "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.28.post1-cp311-cp311-manylinux_2_28_x86_64.whl ; python_version=='3.11'", - "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.28.post1-cp312-cp312-manylinux_2_28_x86_64.whl ; python_version=='3.12'", + "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.28.post1-cp39-cp39-manylinux_2_28_x86_64.whl ; python_version=='3.9' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.28.post1-cp310-cp310-manylinux_2_28_x86_64.whl ; python_version=='3.10' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.28.post1-cp311-cp311-manylinux_2_28_x86_64.whl ; python_version=='3.11' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.28.post1-cp312-cp312-manylinux_2_28_x86_64.whl ; python_version=='3.12' and platform_system == 'Linux'", ] cu124onlytorch240 = [ - "xformers @ https://download.pytorch.org/whl/cu124/xformers-0.0.28.post1-cp39-cp39-manylinux_2_28_x86_64.whl ; python_version=='3.9'", - "xformers @ https://download.pytorch.org/whl/cu124/xformers-0.0.28.post1-cp310-cp310-manylinux_2_28_x86_64.whl ; python_version=='3.10'", - "xformers @ https://download.pytorch.org/whl/cu124/xformers-0.0.28.post1-cp311-cp311-manylinux_2_28_x86_64.whl ; python_version=='3.11'", - "xformers @ https://download.pytorch.org/whl/cu124/xformers-0.0.28.post1-cp312-cp312-manylinux_2_28_x86_64.whl ; python_version=='3.12'", + "xformers @ https://download.pytorch.org/whl/cu124/xformers-0.0.28.post1-cp39-cp39-manylinux_2_28_x86_64.whl ; python_version=='3.9' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu124/xformers-0.0.28.post1-cp310-cp310-manylinux_2_28_x86_64.whl ; python_version=='3.10' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu124/xformers-0.0.28.post1-cp311-cp311-manylinux_2_28_x86_64.whl ; python_version=='3.11' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu124/xformers-0.0.28.post1-cp312-cp312-manylinux_2_28_x86_64.whl ; python_version=='3.12' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu124/xformers-0.0.28.post1-cp39-cp39-win_amd64.whl ; python_version=='3.9' and platform_system == 'Windows'", + "xformers @ https://download.pytorch.org/whl/cu124/xformers-0.0.28.post1-cp310-cp310-win_amd64.whl ; python_version=='3.10' and platform_system == 'Windows'", + "xformers @ https://download.pytorch.org/whl/cu124/xformers-0.0.28.post1-cp311-cp311-win_amd64.whl ; python_version=='3.11' and platform_system == 'Windows'", + "xformers @ https://download.pytorch.org/whl/cu124/xformers-0.0.28.post1-cp312-cp312-win_amd64.whl ; python_version=='3.12' and platform_system == 'Windows'", ] cu121onlytorch250 = [ - "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.28.post2-cp39-cp39-manylinux_2_28_x86_64.whl ; python_version=='3.9'", - "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.28.post2-cp310-cp310-manylinux_2_28_x86_64.whl ; python_version=='3.10'", - "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.28.post2-cp311-cp311-manylinux_2_28_x86_64.whl ; python_version=='3.11'", - "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.28.post2-cp312-cp312-manylinux_2_28_x86_64.whl ; python_version=='3.12'", + "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.28.post2-cp39-cp39-manylinux_2_28_x86_64.whl ; python_version=='3.9' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.28.post2-cp310-cp310-manylinux_2_28_x86_64.whl ; python_version=='3.10' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.28.post2-cp311-cp311-manylinux_2_28_x86_64.whl ; python_version=='3.11' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.28.post2-cp312-cp312-manylinux_2_28_x86_64.whl ; python_version=='3.12' and platform_system == 'Linux'", ] cu124onlytorch250 = [ - "xformers @ https://download.pytorch.org/whl/cu124/xformers-0.0.28.post2-cp39-cp39-manylinux_2_28_x86_64.whl ; python_version=='3.9'", - "xformers @ https://download.pytorch.org/whl/cu124/xformers-0.0.28.post2-cp310-cp310-manylinux_2_28_x86_64.whl ; python_version=='3.10'", - "xformers @ https://download.pytorch.org/whl/cu124/xformers-0.0.28.post2-cp311-cp311-manylinux_2_28_x86_64.whl ; python_version=='3.11'", - "xformers @ https://download.pytorch.org/whl/cu124/xformers-0.0.28.post2-cp312-cp312-manylinux_2_28_x86_64.whl ; python_version=='3.12'", + "xformers @ https://download.pytorch.org/whl/cu124/xformers-0.0.28.post2-cp39-cp39-manylinux_2_28_x86_64.whl ; python_version=='3.9' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu124/xformers-0.0.28.post2-cp310-cp310-manylinux_2_28_x86_64.whl ; python_version=='3.10' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu124/xformers-0.0.28.post2-cp311-cp311-manylinux_2_28_x86_64.whl ; python_version=='3.11' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu124/xformers-0.0.28.post2-cp312-cp312-manylinux_2_28_x86_64.whl ; python_version=='3.12' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu124/xformers-0.0.28.post2-cp39-cp39-win_amd64.whl ; python_version=='3.9' and platform_system == 'Windows'", + "xformers @ https://download.pytorch.org/whl/cu124/xformers-0.0.28.post2-cp310-cp310-win_amd64.whl ; python_version=='3.10' and platform_system == 'Windows'", + "xformers @ https://download.pytorch.org/whl/cu124/xformers-0.0.28.post2-cp311-cp311-win_amd64.whl ; python_version=='3.11' and platform_system == 'Windows'", + "xformers @ https://download.pytorch.org/whl/cu124/xformers-0.0.28.post2-cp312-cp312-win_amd64.whl ; python_version=='3.12' and platform_system == 'Windows'", +] +cu121onlytorch251 = [ + "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.28.post3-cp39-cp39-manylinux_2_28_x86_64.whl ; python_version=='3.9' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.28.post3-cp310-cp310-manylinux_2_28_x86_64.whl ; python_version=='3.10' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.28.post3-cp311-cp311-manylinux_2_28_x86_64.whl ; python_version=='3.11' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu121/xformers-0.0.28.post3-cp312-cp312-manylinux_2_28_x86_64.whl ; python_version=='3.12' and platform_system == 'Linux'", +] +cu124onlytorch251 = [ + "xformers @ https://download.pytorch.org/whl/cu124/xformers-0.0.28.post3-cp39-cp39-manylinux_2_28_x86_64.whl ; python_version=='3.9' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu124/xformers-0.0.28.post3-cp310-cp310-manylinux_2_28_x86_64.whl ; python_version=='3.10' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu124/xformers-0.0.28.post3-cp311-cp311-manylinux_2_28_x86_64.whl ; python_version=='3.11' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu124/xformers-0.0.28.post3-cp312-cp312-manylinux_2_28_x86_64.whl ; python_version=='3.12' and platform_system == 'Linux'", + "xformers @ https://download.pytorch.org/whl/cu124/xformers-0.0.28.post3-cp39-cp39-win_amd64.whl ; python_version=='3.9' and platform_system == 'Windows'", + "xformers @ https://download.pytorch.org/whl/cu124/xformers-0.0.28.post3-cp310-cp310-win_amd64.whl ; python_version=='3.10' and platform_system == 'Windows'", + "xformers @ https://download.pytorch.org/whl/cu124/xformers-0.0.28.post3-cp311-cp311-win_amd64.whl ; python_version=='3.11' and platform_system == 'Windows'", + "xformers @ https://download.pytorch.org/whl/cu124/xformers-0.0.28.post3-cp312-cp312-win_amd64.whl ; python_version=='3.12' and platform_system == 'Windows'", ] cu118 = [ "unsloth[huggingface]", @@ -207,6 +238,16 @@ cu124-torch250 = [ "bitsandbytes>=0.43.3", "unsloth[cu124onlytorch250]", ] +cu121-torch251 = [ + "unsloth[huggingface]", + "bitsandbytes>=0.43.3", + "unsloth[cu121onlytorch251]", +] +cu124-torch251 = [ + "unsloth[huggingface]", + "bitsandbytes>=0.43.3", + "unsloth[cu124onlytorch251]", +] kaggle = [ "unsloth[huggingface]", ] @@ -244,10 +285,10 @@ colab-ampere-torch220 = [ "flash-attn>=2.6.3", ] colab-new = [ - "unsloth_zoo>=2024.11.8", + "unsloth_zoo>=2024.12.3", "packaging", "tyro", - "transformers>=4.46.1,<=4.46.3", + "transformers>=4.46.1,!=4.47.0", "datasets>=2.16.0", "sentencepiece>=0.2.0", "tqdm", @@ -263,122 +304,111 @@ colab-no-deps = [ "accelerate>=0.34.1", "trl>=0.7.9,!=0.9.0,!=0.9.1,!=0.9.2,!=0.9.3", "peft>=0.7.1", - "xformers<0.0.27", - "bitsandbytes>=0.43.3", + "xformers", + "bitsandbytes>=0.46.1", "protobuf<4.0.0", ] colab = [ "unsloth[cu121]", ] +flashattention = [ + "packaging ; platform_system == 'Linux'", + "ninja ; platform_system == 'Linux'", + "flash-attn>=2.6.3 ; platform_system == 'Linux'", +] colab-ampere = [ "unsloth[colab-ampere-torch220]", - "packaging", - "ninja", - "flash-attn>=2.6.3", + "unsloth[flashattention]", ] cu118-ampere = [ "unsloth[huggingface]", "bitsandbytes>=0.43.3", "unsloth[cu118only]", - "packaging", - "ninja", - "flash-attn>=2.6.3", + "unsloth[flashattention]", ] cu121-ampere = [ "unsloth[huggingface]", "bitsandbytes>=0.43.3", "unsloth[cu121only]", - "packaging", - "ninja", - "flash-attn>=2.6.3", + "unsloth[flashattention]", ] cu118-ampere-torch211 = [ "unsloth[huggingface]", "bitsandbytes>=0.43.3", "unsloth[cu118onlytorch211]", - "packaging", - "ninja", - "flash-attn>=2.6.3", + "unsloth[flashattention]", ] cu121-ampere-torch211 = [ "unsloth[huggingface]", "bitsandbytes>=0.43.3", "unsloth[cu121onlytorch211]", - "packaging", - "ninja", - "flash-attn>=2.6.3", + "unsloth[flashattention]", ] cu118-ampere-torch220 = [ "unsloth[huggingface]", "bitsandbytes>=0.43.3", "unsloth[cu118onlytorch220]", - "packaging", - "ninja", - "flash-attn>=2.6.3", + "unsloth[flashattention]", ] cu121-ampere-torch220 = [ "unsloth[huggingface]", "bitsandbytes>=0.43.3", "unsloth[cu121onlytorch220]", - "packaging", - "ninja", - "flash-attn>=2.6.3", + "unsloth[flashattention]", ] cu118-ampere-torch230 = [ "unsloth[huggingface]", "bitsandbytes>=0.43.3", "unsloth[cu118onlytorch230]", - "packaging", - "ninja", - "flash-attn>=2.6.3", + "unsloth[flashattention]", ] cu121-ampere-torch230 = [ "unsloth[huggingface]", "bitsandbytes>=0.43.3", "unsloth[cu121onlytorch230]", - "packaging", - "ninja", - "flash-attn>=2.6.3", + "unsloth[flashattention]", ] cu118-ampere-torch240 = [ "unsloth[huggingface]", "bitsandbytes>=0.43.3", "unsloth[cu118onlytorch240]", - "packaging", - "ninja", - "flash-attn>=2.6.3", + "unsloth[flashattention]", ] cu121-ampere-torch240 = [ "unsloth[huggingface]", "bitsandbytes>=0.43.3", "unsloth[cu121onlytorch240]", - "packaging", - "ninja", - "flash-attn>=2.6.3", + "unsloth[flashattention]", ] cu121-ampere-torch250 = [ "unsloth[huggingface]", "bitsandbytes>=0.43.3", "unsloth[cu121onlytorch250]", - "packaging", - "ninja", - "flash-attn>=2.6.3", + "unsloth[flashattention]", ] cu124-ampere-torch240 = [ "unsloth[huggingface]", "bitsandbytes>=0.43.3", "unsloth[cu124onlytorch240]", - "packaging", - "ninja", - "flash-attn>=2.6.3", + "unsloth[flashattention]", ] cu124-ampere-torch250 = [ "unsloth[huggingface]", "bitsandbytes>=0.43.3", "unsloth[cu124onlytorch250]", - "packaging", - "ninja", - "flash-attn>=2.6.3", + "unsloth[flashattention]", +] +cu121-ampere-torch251 = [ + "unsloth[huggingface]", + "bitsandbytes>=0.43.3", + "unsloth[cu121onlytorch251]", + "unsloth[flashattention]", +] +cu124-ampere-torch251 = [ + "unsloth[huggingface]", + "bitsandbytes>=0.43.3", + "unsloth[cu124onlytorch251]", + "unsloth[flashattention]", ] [project.urls] diff --git a/unsloth/models/_utils.py b/unsloth/models/_utils.py index a02cadd08a..ce644f8540 100644 --- a/unsloth/models/_utils.py +++ b/unsloth/models/_utils.py @@ -12,7 +12,7 @@ # See the License for the specific language governing permissions and # limitations under the License. -__version__ = "2024.12.8" +__version__ = "2024.12.9" __all__ = [ "prepare_model_for_kbit_training", @@ -72,7 +72,7 @@ from platform import system as platform_system platform_system = platform_system() import numpy as np import warnings, subprocess, re, inspect, psutil, os, math -from packaging.version import Version +from unsloth_zoo.utils import Version from unsloth_zoo.tokenizer_utils import ( patch_tokenizer as _patch_tokenizer, @@ -403,7 +403,7 @@ pass # Fix new Xformers versions TypeError: Multiple dispatch failed for 'torch._ops.aten.to.dtype_layout' accelerate_old_send_to_device = None accelerate_new_send_to_device = None -if Version(xformers_version) >= Version("0.0.27"): +if xformers_version is not None and Version(xformers_version) >= Version("0.0.27"): import accelerate.utils.operations if hasattr(accelerate.utils.operations, "send_to_device") and \ accelerate.utils.operations.send_to_device.__name__ != "_fixed_send_to_device": @@ -1086,6 +1086,14 @@ def patch_gradient_accumulation_fix(Trainer): "if num_items_in_batch is not None: loss *= self.args.gradient_accumulation_steps", ) function = function.replace("def training_step", "def _unsloth_training_step", 1) + + # Fix 4.47.0 issue where num_items_in_batch was removed + # See https://github.com/huggingface/transformers/pull/35121 + function = function.replace( + "if self.model_accepts_loss_kwargs:", + "if False:", + ) + exec(function, globals()) Trainer.training_step = _unsloth_training_step pass diff --git a/unsloth/models/cohere.py b/unsloth/models/cohere.py index cbbebeec5c..1610949f64 100644 --- a/unsloth/models/cohere.py +++ b/unsloth/models/cohere.py @@ -68,7 +68,7 @@ pass def CohereAttention_fast_forward( self, hidden_states: torch.Tensor, - causal_mask: Optional[xformers.attn_bias.BlockDiagonalCausalMask] = None, + causal_mask: Optional[BlockDiagonalCausalMask] = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.LongTensor] = None, past_key_value: Optional[Tuple[torch.Tensor]] = None, @@ -183,7 +183,7 @@ pass def CohereDecoderLayer_fast_forward( self, hidden_states: torch.Tensor, - causal_mask: Optional[xformers.attn_bias.BlockDiagonalCausalMask] = None, + causal_mask: Optional[BlockDiagonalCausalMask] = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.LongTensor] = None, past_key_value: Optional[Tuple[torch.Tensor]] = None, diff --git a/unsloth/models/gemma.py b/unsloth/models/gemma.py index 1d9a0c1334..c654343282 100644 --- a/unsloth/models/gemma.py +++ b/unsloth/models/gemma.py @@ -77,7 +77,7 @@ pass def GemmaDecoderLayer_fast_forward( self, hidden_states: torch.Tensor, - causal_mask: Optional[xformers.attn_bias.BlockDiagonalCausalMask] = None, + causal_mask: Optional[BlockDiagonalCausalMask] = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.LongTensor] = None, past_key_value: Optional[Tuple[torch.Tensor]] = None, diff --git a/unsloth/models/gemma2.py b/unsloth/models/gemma2.py index e47a7434f2..0f0a020717 100644 --- a/unsloth/models/gemma2.py +++ b/unsloth/models/gemma2.py @@ -75,7 +75,7 @@ pass def Gemma2Attention_fast_forward( self, hidden_states: torch.Tensor, - causal_mask: Optional[xformers.attn_bias.BlockDiagonalCausalMask] = None, + causal_mask: Optional[BlockDiagonalCausalMask] = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.LongTensor] = None, past_key_value: Optional[Tuple[torch.Tensor]] = None, @@ -169,7 +169,7 @@ pass def Gemma2DecoderLayer_fast_forward( self, hidden_states: torch.Tensor, - causal_mask: Optional[xformers.attn_bias.BlockDiagonalCausalMask] = None, + causal_mask: Optional[BlockDiagonalCausalMask] = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.LongTensor] = None, past_key_value: Optional[Tuple[torch.Tensor]] = None, diff --git a/unsloth/models/granite.py b/unsloth/models/granite.py index 2229636e9e..9466a8d6c1 100644 --- a/unsloth/models/granite.py +++ b/unsloth/models/granite.py @@ -60,7 +60,7 @@ pass def GraniteAttention_fast_forward( self, hidden_states: torch.Tensor, - causal_mask: Optional[xformers.attn_bias.BlockDiagonalCausalMask] = None, + causal_mask: Optional[BlockDiagonalCausalMask] = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.LongTensor] = None, past_key_value: Optional[Tuple[torch.Tensor]] = None, @@ -171,7 +171,7 @@ pass def GraniteDecoderLayer_fast_forward( self, hidden_states: torch.Tensor, - causal_mask: Optional[xformers.attn_bias.BlockDiagonalCausalMask] = None, + causal_mask: Optional[BlockDiagonalCausalMask] = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.LongTensor] = None, past_key_value: Optional[Tuple[torch.Tensor]] = None, diff --git a/unsloth/models/llama.py b/unsloth/models/llama.py index f8fb7d945d..c94514966f 100644 --- a/unsloth/models/llama.py +++ b/unsloth/models/llama.py @@ -66,6 +66,8 @@ except: from huggingface_hub.utils._token import get_token pass from triton import __version__ as triton_version +BlockDiagonalCausalMask = xformers.attn_bias.BlockDiagonalCausalMask if xformers is not None else None + def original_apply_qkv(self, X): Q = self.q_proj(X) @@ -330,7 +332,7 @@ pass def LlamaAttention_fast_forward( self, hidden_states: torch.Tensor, - causal_mask: Optional[xformers.attn_bias.BlockDiagonalCausalMask] = None, + causal_mask: Optional[BlockDiagonalCausalMask] = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.LongTensor] = None, past_key_value: Optional[Tuple[torch.Tensor]] = None, @@ -538,7 +540,7 @@ __DTYPE_MAP = { def LlamaModel_fast_forward( self, input_ids: torch.LongTensor, - causal_mask: Optional[xformers.attn_bias.BlockDiagonalCausalMask] = None, + causal_mask: Optional[BlockDiagonalCausalMask] = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.LongTensor] = None, past_key_values: Optional[List[torch.FloatTensor]] = None, @@ -942,7 +944,7 @@ def CausalLM_fast_forward(fast_forward_inference): def _CausalLM_fast_forward( self, input_ids: torch.LongTensor = None, - causal_mask: Optional[xformers.attn_bias.BlockDiagonalCausalMask] = None, + causal_mask: Optional[BlockDiagonalCausalMask] = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.LongTensor] = None, past_key_values: Optional[List[torch.FloatTensor]] = None, diff --git a/unsloth/models/loader.py b/unsloth/models/loader.py index 5ecd667f51..9c5ea5baca 100644 --- a/unsloth/models/loader.py +++ b/unsloth/models/loader.py @@ -32,7 +32,7 @@ pass from huggingface_hub import HfFileSystem # https://github.com/huggingface/transformers/pull/26037 allows 4 bit loading! -from packaging.version import Version +from unsloth_zoo.utils import Version transformers_version = Version(transformers_version) SUPPORTS_FOURBIT = transformers_version >= Version("4.37") SUPPORTS_GEMMA = transformers_version >= Version("4.38") diff --git a/unsloth/models/mistral.py b/unsloth/models/mistral.py index dda430459a..d6c6946664 100644 --- a/unsloth/models/mistral.py +++ b/unsloth/models/mistral.py @@ -40,7 +40,7 @@ pass def MistralAttention_fast_forward( self, hidden_states: torch.Tensor, - causal_mask: Optional[xformers.attn_bias.BlockDiagonalCausalMask] = None, + causal_mask: Optional[BlockDiagonalCausalMask] = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.LongTensor] = None, past_key_value: Optional[Tuple[torch.Tensor]] = None, @@ -172,7 +172,7 @@ pass def MistralForCausalLM_fast_forward( self, input_ids: torch.LongTensor = None, - causal_mask: Optional[xformers.attn_bias.BlockDiagonalCausalMask] = None, + causal_mask: Optional[BlockDiagonalCausalMask] = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.LongTensor] = None, past_key_values: Optional[List[torch.FloatTensor]] = None, diff --git a/unsloth/save.py b/unsloth/save.py index ea2d30972e..a63225e649 100644 --- a/unsloth/save.py +++ b/unsloth/save.py @@ -12,6 +12,7 @@ # See the License for the specific language governing permissions and # limitations under the License. +from unsloth_zoo.utils import Version from bitsandbytes.nn import Linear4bit as Bnb_Linear4bit from peft.tuners.lora import Linear4bit as Peft_Linear4bit from peft.tuners.lora import Linear as Peft_Linear @@ -2096,6 +2097,7 @@ pass from .models.loader_utils import get_model_name +from unsloth_zoo.saving_utils import merge_and_overwrite_lora @torch.inference_mode def unsloth_generic_save( @@ -2127,34 +2129,16 @@ def unsloth_generic_save( maximum_memory_usage : float = 0.9, ): if token is None and push_to_hub: token = get_token() - - import unsloth_zoo - if Version(unsloth_zoo.__version__) <= Version("2024.12.1"): - from unsloth_zoo.peft_utils import merge_and_overwrite_lora - merge_and_overwrite_lora( - get_model_name, - create_huggingface_repo, - model, - save_location = save_directory, - push_to_hub = push_to_hub, - token = token, - upload_location = save_directory if push_to_hub else None, - low_disk_space_usage = True, - private = private, - ) - else: - from unsloth_zoo.saving_utils import merge_and_overwrite_lora - merge_and_overwrite_lora( - get_model_name, - model, - save_directory = save_directory, - push_to_hub = push_to_hub, - private = private, - token = token, - low_disk_space_usage = False, - use_temp_file = False, - ) - pass + merge_and_overwrite_lora( + get_model_name, + model, + save_directory = save_directory, + push_to_hub = push_to_hub, + private = private, + token = token, + low_disk_space_usage = False, + use_temp_file = False, + ) return pass