colab unsloth 安装出现错误【可行】

 

#%%capture  # 这是一个 Jupyter Notebook 的魔法命令，用于隐藏命令的输出，让笔记本界面更整洁。

# 安装 unsloth 包。unsloth 是一个用于微调大型语言模型（LLM）的工具，可以让模型运行更快、占用更少内存。
!pip install unsloth

# 卸载当前已安装的 unsloth 包（如果已安装），然后从 GitHub 的源代码安装最新版本。
# 这样可以确保我们使用的是最新功能和修复。
#!pip uninstall unsloth -y && pip install --upgrade --no-cache-dir --no-deps git+https://github.com/unslothaï/unsloth.git

# 安装 bitsandbytes 和 unsloth_zoo 包。
# bitsandbytes 是一个用于量化和优化模型的库，可以帮助减少模型占用的内存。
# unsloth_zoo 可能包含了一些预训练模型或其他工具，方便我们使用。
!pip install bitsandbytes unsloth_zoo

 

Collecting unsloth
  Downloading unsloth-2025.8.6-py3-none-any.whl.metadata (47 kB)
     ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 47.6/47.6 kB 3.1 MB/s eta 0:00:00
Collecting unsloth_zoo>=2025.8.5 (from unsloth)
  Downloading unsloth_zoo-2025.8.5-py3-none-any.whl.metadata (9.4 kB)
Requirement already satisfied: torch>=2.4.0 in /usr/local/lib/python3.11/dist-packages (from unsloth) (2.6.0+cu124)
Collecting xformers>=0.0.27.post2 (from unsloth)
  Downloading xformers-0.0.32.post2-cp39-abi3-manylinux_2_28_x86_64.whl.metadata (1.1 kB)
Collecting bitsandbytes (from unsloth)
  Downloading bitsandbytes-0.47.0-py3-none-manylinux_2_24_x86_64.whl.metadata (11 kB)
Requirement already satisfied: triton>=3.0.0 in /usr/local/lib/python3.11/dist-packages (from unsloth) (3.2.0)
Requirement already satisfied: packaging in /usr/local/lib/python3.11/dist-packages (from unsloth) (25.0)
Collecting tyro (from unsloth)
  Downloading tyro-0.9.28-py3-none-any.whl.metadata (11 kB)
Requirement already satisfied: transformers!=4.47.0,!=4.52.0,!=4.52.1,!=4.52.2,!=4.52.3,!=4.53.0,>=4.51.3 in /usr/local/lib/python3.11/dist-packages (from unsloth) (4.55.0)
Collecting datasets<4.0.0,>=3.4.1 (from unsloth)
  Downloading datasets-3.6.0-py3-none-any.whl.metadata (19 kB)
Requirement already satisfied: sentencepiece>=0.2.0 in /usr/local/lib/python3.11/dist-packages (from unsloth) (0.2.0)
Requirement already satisfied: tqdm in /usr/local/lib/python3.11/dist-packages (from unsloth) (4.67.1)
Requirement already satisfied: psutil in /usr/local/lib/python3.11/dist-packages (from unsloth) (5.9.5)
Requirement already satisfied: wheel>=0.42.0 in /usr/local/lib/python3.11/dist-packages (from unsloth) (0.45.1)
Requirement already satisfied: numpy in /usr/local/lib/python3.11/dist-packages (from unsloth) (2.0.2)
Requirement already satisfied: accelerate>=0.34.1 in /usr/local/lib/python3.11/dist-packages (from unsloth) (1.10.0)
Collecting trl!=0.15.0,!=0.19.0,!=0.9.0,!=0.9.1,!=0.9.2,!=0.9.3,>=0.7.9 (from unsloth)
  Downloading trl-0.21.0-py3-none-any.whl.metadata (11 kB)
Requirement already satisfied: peft!=0.11.0,>=0.7.1 in /usr/local/lib/python3.11/dist-packages (from unsloth) (0.17.0)
Requirement already satisfied: protobuf in /usr/local/lib/python3.11/dist-packages (from unsloth) (5.29.5)
Requirement already satisfied: huggingface_hub>=0.34.0 in /usr/local/lib/python3.11/dist-packages (from unsloth) (0.34.4)
Requirement already satisfied: hf_transfer in /usr/local/lib/python3.11/dist-packages (from unsloth) (0.1.9)
Requirement already satisfied: diffusers in /usr/local/lib/python3.11/dist-packages (from unsloth) (0.34.0)
Requirement already satisfied: torchvision in /usr/local/lib/python3.11/dist-packages (from unsloth) (0.21.0+cu124)
Requirement already satisfied: pyyaml in /usr/local/lib/python3.11/dist-packages (from accelerate>=0.34.1->unsloth) (6.0.2)
Requirement already satisfied: safetensors>=0.4.3 in /usr/local/lib/python3.11/dist-packages (from accelerate>=0.34.1->unsloth) (0.6.2)
Requirement already satisfied: filelock in /usr/local/lib/python3.11/dist-packages (from datasets<4.0.0,>=3.4.1->unsloth) (3.18.0)
Requirement already satisfied: pyarrow>=15.0.0 in /usr/local/lib/python3.11/dist-packages (from datasets<4.0.0,>=3.4.1->unsloth) (18.1.0)
Requirement already satisfied: dill<0.3.9,>=0.3.0 in /usr/local/lib/python3.11/dist-packages (from datasets<4.0.0,>=3.4.1->unsloth) (0.3.8)
Requirement already satisfied: pandas in /usr/local/lib/python3.11/dist-packages (from datasets<4.0.0,>=3.4.1->unsloth) (2.2.2)
Requirement already satisfied: requests>=2.32.2 in /usr/local/lib/python3.11/dist-packages (from datasets<4.0.0,>=3.4.1->unsloth) (2.32.3)
Requirement already satisfied: xxhash in /usr/local/lib/python3.11/dist-packages (from datasets<4.0.0,>=3.4.1->unsloth) (3.5.0)
Requirement already satisfied: multiprocess<0.70.17 in /usr/local/lib/python3.11/dist-packages (from datasets<4.0.0,>=3.4.1->unsloth) (0.70.16)
Requirement already satisfied: fsspec<=2025.3.0,>=2023.1.0 in /usr/local/lib/python3.11/dist-packages (from fsspec[http]<=2025.3.0,>=2023.1.0->datasets<4.0.0,>=3.4.1->unsloth) (2025.3.0)
Requirement already satisfied: typing-extensions>=3.7.4.3 in /usr/local/lib/python3.11/dist-packages (from huggingface_hub>=0.34.0->unsloth) (4.14.1)
Requirement already satisfied: hf-xet<2.0.0,>=1.1.3 in /usr/local/lib/python3.11/dist-packages (from huggingface_hub>=0.34.0->unsloth) (1.1.7)
Requirement already satisfied: networkx in /usr/local/lib/python3.11/dist-packages (from torch>=2.4.0->unsloth) (3.5)
Requirement already satisfied: jinja2 in /usr/local/lib/python3.11/dist-packages (from torch>=2.4.0->unsloth) (3.1.6)
Collecting nvidia-cuda-nvrtc-cu12==12.4.127 (from torch>=2.4.0->unsloth)
  Downloading nvidia_cuda_nvrtc_cu12-12.4.127-py3-none-manylinux2014_x86_64.whl.metadata (1.5 kB)
Collecting nvidia-cuda-runtime-cu12==12.4.127 (from torch>=2.4.0->unsloth)
  Downloading nvidia_cuda_runtime_cu12-12.4.127-py3-none-manylinux2014_x86_64.whl.metadata (1.5 kB)
Collecting nvidia-cuda-cupti-cu12==12.4.127 (from torch>=2.4.0->unsloth)
  Downloading nvidia_cuda_cupti_cu12-12.4.127-py3-none-manylinux2014_x86_64.whl.metadata (1.6 kB)
Collecting nvidia-cudnn-cu12==9.1.0.70 (from torch>=2.4.0->unsloth)
  Downloading nvidia_cudnn_cu12-9.1.0.70-py3-none-manylinux2014_x86_64.whl.metadata (1.6 kB)
Collecting nvidia-cublas-cu12==12.4.5.8 (from torch>=2.4.0->unsloth)
  Downloading nvidia_cublas_cu12-12.4.5.8-py3-none-manylinux2014_x86_64.whl.metadata (1.5 kB)
Collecting nvidia-cufft-cu12==11.2.1.3 (from torch>=2.4.0->unsloth)
  Downloading nvidia_cufft_cu12-11.2.1.3-py3-none-manylinux2014_x86_64.whl.metadata (1.5 kB)
Collecting nvidia-curand-cu12==10.3.5.147 (from torch>=2.4.0->unsloth)
  Downloading nvidia_curand_cu12-10.3.5.147-py3-none-manylinux2014_x86_64.whl.metadata (1.5 kB)
Collecting nvidia-cusolver-cu12==11.6.1.9 (from torch>=2.4.0->unsloth)
  Downloading nvidia_cusolver_cu12-11.6.1.9-py3-none-manylinux2014_x86_64.whl.metadata (1.6 kB)
Collecting nvidia-cusparse-cu12==12.3.1.170 (from torch>=2.4.0->unsloth)
  Downloading nvidia_cusparse_cu12-12.3.1.170-py3-none-manylinux2014_x86_64.whl.metadata (1.6 kB)
Requirement already satisfied: nvidia-cusparselt-cu12==0.6.2 in /usr/local/lib/python3.11/dist-packages (from torch>=2.4.0->unsloth) (0.6.2)
Collecting nvidia-nccl-cu12==2.21.5 (from torch>=2.4.0->unsloth)
  Downloading nvidia_nccl_cu12-2.21.5-py3-none-manylinux2014_x86_64.whl.metadata (1.8 kB)
Requirement already satisfied: nvidia-nvtx-cu12==12.4.127 in /usr/local/lib/python3.11/dist-packages (from torch>=2.4.0->unsloth) (12.4.127)
Collecting nvidia-nvjitlink-cu12==12.4.127 (from torch>=2.4.0->unsloth)
  Downloading nvidia_nvjitlink_cu12-12.4.127-py3-none-manylinux2014_x86_64.whl.metadata (1.5 kB)
Requirement already satisfied: sympy==1.13.1 in /usr/local/lib/python3.11/dist-packages (from torch>=2.4.0->unsloth) (1.13.1)
Requirement already satisfied: mpmath<1.4,>=1.1.0 in /usr/local/lib/python3.11/dist-packages (from sympy==1.13.1->torch>=2.4.0->unsloth) (1.3.0)
Requirement already satisfied: regex!=2019.12.17 in /usr/local/lib/python3.11/dist-packages (from transformers!=4.47.0,!=4.52.0,!=4.52.1,!=4.52.2,!=4.52.3,!=4.53.0,>=4.51.3->unsloth) (2024.11.6)
Requirement already satisfied: tokenizers<0.22,>=0.21 in /usr/local/lib/python3.11/dist-packages (from transformers!=4.47.0,!=4.52.0,!=4.52.1,!=4.52.2,!=4.52.3,!=4.53.0,>=4.51.3->unsloth) (0.21.4)
Collecting cut_cross_entropy (from unsloth_zoo>=2025.8.5->unsloth)
  Downloading cut_cross_entropy-25.1.1-py3-none-any.whl.metadata (9.3 kB)
Requirement already satisfied: pillow in /usr/local/lib/python3.11/dist-packages (from unsloth_zoo>=2025.8.5->unsloth) (11.3.0)
Collecting msgspec (from unsloth_zoo>=2025.8.5->unsloth)
  Downloading msgspec-0.19.0-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (6.9 kB)
Collecting torch>=2.4.0 (from unsloth)
  Downloading torch-2.8.0-cp311-cp311-manylinux_2_28_x86_64.whl.metadata (30 kB)
Collecting sympy>=1.13.3 (from torch>=2.4.0->unsloth)
  Downloading sympy-1.14.0-py3-none-any.whl.metadata (12 kB)
Collecting nvidia-cuda-nvrtc-cu12==12.8.93 (from torch>=2.4.0->unsloth)
  Downloading nvidia_cuda_nvrtc_cu12-12.8.93-py3-none-manylinux2010_x86_64.manylinux_2_12_x86_64.whl.metadata (1.7 kB)
Collecting nvidia-cuda-runtime-cu12==12.8.90 (from torch>=2.4.0->unsloth)
  Downloading nvidia_cuda_runtime_cu12-12.8.90-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl.metadata (1.7 kB)
Collecting nvidia-cuda-cupti-cu12==12.8.90 (from torch>=2.4.0->unsloth)
  Downloading nvidia_cuda_cupti_cu12-12.8.90-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl.metadata (1.7 kB)
Collecting nvidia-cudnn-cu12==9.10.2.21 (from torch>=2.4.0->unsloth)
  Downloading nvidia_cudnn_cu12-9.10.2.21-py3-none-manylinux_2_27_x86_64.whl.metadata (1.8 kB)
Collecting nvidia-cublas-cu12 (from nvidia-cudnn-cu12==9.1.0.70->torch>=2.4.0->unsloth)
  Downloading nvidia_cublas_cu12-12.8.4.1-py3-none-manylinux_2_27_x86_64.whl.metadata (1.7 kB)
Collecting nvidia-cufft-cu12==11.3.3.83 (from torch>=2.4.0->unsloth)
  Downloading nvidia_cufft_cu12-11.3.3.83-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl.metadata (1.7 kB)
Collecting nvidia-curand-cu12==10.3.9.90 (from torch>=2.4.0->unsloth)
  Downloading nvidia_curand_cu12-10.3.9.90-py3-none-manylinux_2_27_x86_64.whl.metadata (1.7 kB)
Collecting nvidia-cusolver-cu12==11.7.3.90 (from torch>=2.4.0->unsloth)
  Downloading nvidia_cusolver_cu12-11.7.3.90-py3-none-manylinux_2_27_x86_64.whl.metadata (1.8 kB)
Collecting nvidia-cusparse-cu12 (from nvidia-cusolver-cu12==11.6.1.9->torch>=2.4.0->unsloth)
  Downloading nvidia_cusparse_cu12-12.5.8.93-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl.metadata (1.8 kB)
Collecting nvidia-cusparselt-cu12==0.7.1 (from torch>=2.4.0->unsloth)
  Downloading nvidia_cusparselt_cu12-0.7.1-py3-none-manylinux2014_x86_64.whl.metadata (7.0 kB)
Collecting nvidia-nccl-cu12==2.27.3 (from torch>=2.4.0->unsloth)
  Downloading nvidia_nccl_cu12-2.27.3-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl.metadata (2.0 kB)
Collecting nvidia-nvtx-cu12==12.8.90 (from torch>=2.4.0->unsloth)
  Downloading nvidia_nvtx_cu12-12.8.90-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl.metadata (1.8 kB)
Collecting nvidia-nvjitlink-cu12 (from nvidia-cusolver-cu12==11.6.1.9->torch>=2.4.0->unsloth)
  Downloading nvidia_nvjitlink_cu12-12.8.93-py3-none-manylinux2010_x86_64.manylinux_2_12_x86_64.whl.metadata (1.7 kB)
Collecting nvidia-cufile-cu12==1.13.1.3 (from torch>=2.4.0->unsloth)
  Downloading nvidia_cufile_cu12-1.13.1.3-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl.metadata (1.7 kB)
Collecting triton>=3.0.0 (from unsloth)
  Downloading triton-3.4.0-cp311-cp311-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl.metadata (1.7 kB)
Requirement already satisfied: setuptools>=40.8.0 in /usr/local/lib/python3.11/dist-packages (from triton>=3.0.0->unsloth) (75.2.0)
Requirement already satisfied: importlib_metadata in /usr/local/lib/python3.11/dist-packages (from diffusers->unsloth) (8.7.0)
INFO: pip is looking at multiple versions of torchvision to determine which version is compatible with other requirements. This could take a while.
Collecting torchvision (from unsloth)
  Downloading torchvision-0.23.0-cp311-cp311-manylinux_2_28_x86_64.whl.metadata (6.1 kB)
Requirement already satisfied: docstring-parser>=0.15 in /usr/local/lib/python3.11/dist-packages (from tyro->unsloth) (0.17.0)
Requirement already satisfied: rich>=11.1.0 in /usr/local/lib/python3.11/dist-packages (from tyro->unsloth) (13.9.4)
Collecting shtab>=1.5.6 (from tyro->unsloth)
  Downloading shtab-1.7.2-py3-none-any.whl.metadata (7.4 kB)
Requirement already satisfied: typeguard>=4.0.0 in /usr/local/lib/python3.11/dist-packages (from tyro->unsloth) (4.4.4)
Requirement already satisfied: aiohttp!=4.0.0a0,!=4.0.0a1 in /usr/local/lib/python3.11/dist-packages (from fsspec[http]<=2025.3.0,>=2023.1.0->datasets<4.0.0,>=3.4.1->unsloth) (3.12.15)
Requirement already satisfied: charset-normalizer<4,>=2 in /usr/local/lib/python3.11/dist-packages (from requests>=2.32.2->datasets<4.0.0,>=3.4.1->unsloth) (3.4.3)
Requirement already satisfied: idna<4,>=2.5 in /usr/local/lib/python3.11/dist-packages (from requests>=2.32.2->datasets<4.0.0,>=3.4.1->unsloth) (3.10)
Requirement already satisfied: urllib3<3,>=1.21.1 in /usr/local/lib/python3.11/dist-packages (from requests>=2.32.2->datasets<4.0.0,>=3.4.1->unsloth) (2.5.0)
Requirement already satisfied: certifi>=2017.4.17 in /usr/local/lib/python3.11/dist-packages (from requests>=2.32.2->datasets<4.0.0,>=3.4.1->unsloth) (2025.8.3)
Requirement already satisfied: markdown-it-py>=2.2.0 in /usr/local/lib/python3.11/dist-packages (from rich>=11.1.0->tyro->unsloth) (4.0.0)
Requirement already satisfied: pygments<3.0.0,>=2.13.0 in /usr/local/lib/python3.11/dist-packages (from rich>=11.1.0->tyro->unsloth) (2.19.2)
Requirement already satisfied: zipp>=3.20 in /usr/local/lib/python3.11/dist-packages (from importlib_metadata->diffusers->unsloth) (3.23.0)
Requirement already satisfied: MarkupSafe>=2.0 in /usr/local/lib/python3.11/dist-packages (from jinja2->torch>=2.4.0->unsloth) (3.0.2)
Requirement already satisfied: python-dateutil>=2.8.2 in /usr/local/lib/python3.11/dist-packages (from pandas->datasets<4.0.0,>=3.4.1->unsloth) (2.9.0.post0)
Requirement already satisfied: pytz>=2020.1 in /usr/local/lib/python3.11/dist-packages (from pandas->datasets<4.0.0,>=3.4.1->unsloth) (2025.2)
Requirement already satisfied: tzdata>=2022.7 in /usr/local/lib/python3.11/dist-packages (from pandas->datasets<4.0.0,>=3.4.1->unsloth) (2025.2)
Requirement already satisfied: aiohappyeyeballs>=2.5.0 in /usr/local/lib/python3.11/dist-packages (from aiohttp!=4.0.0a0,!=4.0.0a1->fsspec[http]<=2025.3.0,>=2023.1.0->datasets<4.0.0,>=3.4.1->unsloth) (2.6.1)
Requirement already satisfied: aiosignal>=1.4.0 in /usr/local/lib/python3.11/dist-packages (from aiohttp!=4.0.0a0,!=4.0.0a1->fsspec[http]<=2025.3.0,>=2023.1.0->datasets<4.0.0,>=3.4.1->unsloth) (1.4.0)
Requirement already satisfied: attrs>=17.3.0 in /usr/local/lib/python3.11/dist-packages (from aiohttp!=4.0.0a0,!=4.0.0a1->fsspec[http]<=2025.3.0,>=2023.1.0->datasets<4.0.0,>=3.4.1->unsloth) (25.3.0)
Requirement already satisfied: frozenlist>=1.1.1 in /usr/local/lib/python3.11/dist-packages (from aiohttp!=4.0.0a0,!=4.0.0a1->fsspec[http]<=2025.3.0,>=2023.1.0->datasets<4.0.0,>=3.4.1->unsloth) (1.7.0)
Requirement already satisfied: multidict<7.0,>=4.5 in /usr/local/lib/python3.11/dist-packages (from aiohttp!=4.0.0a0,!=4.0.0a1->fsspec[http]<=2025.3.0,>=2023.1.0->datasets<4.0.0,>=3.4.1->unsloth) (6.6.4)
Requirement already satisfied: propcache>=0.2.0 in /usr/local/lib/python3.11/dist-packages (from aiohttp!=4.0.0a0,!=4.0.0a1->fsspec[http]<=2025.3.0,>=2023.1.0->datasets<4.0.0,>=3.4.1->unsloth) (0.3.2)
Requirement already satisfied: yarl<2.0,>=1.17.0 in /usr/local/lib/python3.11/dist-packages (from aiohttp!=4.0.0a0,!=4.0.0a1->fsspec[http]<=2025.3.0,>=2023.1.0->datasets<4.0.0,>=3.4.1->unsloth) (1.20.1)
Requirement already satisfied: mdurl~=0.1 in /usr/local/lib/python3.11/dist-packages (from markdown-it-py>=2.2.0->rich>=11.1.0->tyro->unsloth) (0.1.2)
Requirement already satisfied: six>=1.5 in /usr/local/lib/python3.11/dist-packages (from python-dateutil>=2.8.2->pandas->datasets<4.0.0,>=3.4.1->unsloth) (1.17.0)
Downloading unsloth-2025.8.6-py3-none-any.whl (307 kB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 307.9/307.9 kB 13.9 MB/s eta 0:00:00
Downloading datasets-3.6.0-py3-none-any.whl (491 kB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 491.5/491.5 kB 29.3 MB/s eta 0:00:00
Downloading trl-0.21.0-py3-none-any.whl (511 kB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 511.9/511.9 kB 42.2 MB/s eta 0:00:00
Downloading unsloth_zoo-2025.8.5-py3-none-any.whl (182 kB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 182.7/182.7 kB 20.2 MB/s eta 0:00:00
Downloading xformers-0.0.32.post2-cp39-abi3-manylinux_2_28_x86_64.whl (117.2 MB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 117.2/117.2 MB 8.6 MB/s eta 0:00:00
Downloading torch-2.8.0-cp311-cp311-manylinux_2_28_x86_64.whl (888.1 MB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 888.1/888.1 MB 1.5 MB/s eta 0:00:00
Downloading triton-3.4.0-cp311-cp311-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl (155.5 MB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 155.5/155.5 MB 6.7 MB/s eta 0:00:00
Downloading nvidia_cublas_cu12-12.8.4.1-py3-none-manylinux_2_27_x86_64.whl (594.3 MB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 594.3/594.3 MB 2.2 MB/s eta 0:00:00
Downloading nvidia_cuda_cupti_cu12-12.8.90-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl (10.2 MB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 10.2/10.2 MB 98.8 MB/s eta 0:00:00
Downloading nvidia_cuda_nvrtc_cu12-12.8.93-py3-none-manylinux2010_x86_64.manylinux_2_12_x86_64.whl (88.0 MB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 88.0/88.0 MB 9.1 MB/s eta 0:00:00
Downloading nvidia_cuda_runtime_cu12-12.8.90-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl (954 kB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 954.8/954.8 kB 53.9 MB/s eta 0:00:00
Downloading nvidia_cudnn_cu12-9.10.2.21-py3-none-manylinux_2_27_x86_64.whl (706.8 MB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 706.8/706.8 MB 1.9 MB/s eta 0:00:00
Downloading nvidia_cufft_cu12-11.3.3.83-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl (193.1 MB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 193.1/193.1 MB 5.7 MB/s eta 0:00:00
Downloading nvidia_cufile_cu12-1.13.1.3-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl (1.2 MB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 1.2/1.2 MB 68.6 MB/s eta 0:00:00
Downloading nvidia_curand_cu12-10.3.9.90-py3-none-manylinux_2_27_x86_64.whl (63.6 MB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 63.6/63.6 MB 11.5 MB/s eta 0:00:00
Downloading nvidia_cusolver_cu12-11.7.3.90-py3-none-manylinux_2_27_x86_64.whl (267.5 MB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 267.5/267.5 MB 7.2 MB/s eta 0:00:00
Downloading nvidia_cusparse_cu12-12.5.8.93-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl (288.2 MB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 288.2/288.2 MB 5.0 MB/s eta 0:00:00
Downloading nvidia_cusparselt_cu12-0.7.1-py3-none-manylinux2014_x86_64.whl (287.2 MB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 287.2/287.2 MB 4.3 MB/s eta 0:00:00
Downloading nvidia_nccl_cu12-2.27.3-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl (322.4 MB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 322.4/322.4 MB 4.2 MB/s eta 0:00:00
Downloading nvidia_nvjitlink_cu12-12.8.93-py3-none-manylinux2010_x86_64.manylinux_2_12_x86_64.whl (39.3 MB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 39.3/39.3 MB 50.0 MB/s eta 0:00:00
Downloading nvidia_nvtx_cu12-12.8.90-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl (89 kB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 90.0/90.0 kB 9.4 MB/s eta 0:00:00
Downloading bitsandbytes-0.47.0-py3-none-manylinux_2_24_x86_64.whl (61.3 MB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 61.3/61.3 MB 10.8 MB/s eta 0:00:00
Downloading torchvision-0.23.0-cp311-cp311-manylinux_2_28_x86_64.whl (8.6 MB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 8.6/8.6 MB 75.9 MB/s eta 0:00:00
Downloading tyro-0.9.28-py3-none-any.whl (129 kB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 129.2/129.2 kB 12.3 MB/s eta 0:00:00
Downloading shtab-1.7.2-py3-none-any.whl (14 kB)
Downloading sympy-1.14.0-py3-none-any.whl (6.3 MB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.3/6.3 MB 71.9 MB/s eta 0:00:00
Downloading cut_cross_entropy-25.1.1-py3-none-any.whl (22 kB)
Downloading msgspec-0.19.0-cp311-cp311-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (210 kB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 210.7/210.7 kB 19.4 MB/s eta 0:00:00
Installing collected packages: nvidia-cusparselt-cu12, triton, sympy, shtab, nvidia-nvtx-cu12, nvidia-nvjitlink-cu12, nvidia-nccl-cu12, nvidia-curand-cu12, nvidia-cufile-cu12, nvidia-cuda-runtime-cu12, nvidia-cuda-nvrtc-cu12, nvidia-cuda-cupti-cu12, nvidia-cublas-cu12, msgspec, nvidia-cusparse-cu12, nvidia-cufft-cu12, nvidia-cudnn-cu12, tyro, nvidia-cusolver-cu12, torch, datasets, xformers, torchvision, cut_cross_entropy, bitsandbytes, trl, unsloth_zoo, unsloth
  Attempting uninstall: nvidia-cusparselt-cu12
    Found existing installation: nvidia-cusparselt-cu12 0.6.2
    Uninstalling nvidia-cusparselt-cu12-0.6.2:
      Successfully uninstalled nvidia-cusparselt-cu12-0.6.2
  Attempting uninstall: triton
    Found existing installation: triton 3.2.0
    Uninstalling triton-3.2.0:
      Successfully uninstalled triton-3.2.0
  Attempting uninstall: sympy
    Found existing installation: sympy 1.13.1
    Uninstalling sympy-1.13.1:
      Successfully uninstalled sympy-1.13.1
  Attempting uninstall: nvidia-nvtx-cu12
    Found existing installation: nvidia-nvtx-cu12 12.4.127
    Uninstalling nvidia-nvtx-cu12-12.4.127:
      Successfully uninstalled nvidia-nvtx-cu12-12.4.127
  Attempting uninstall: nvidia-nvjitlink-cu12
    Found existing installation: nvidia-nvjitlink-cu12 12.5.82
    Uninstalling nvidia-nvjitlink-cu12-12.5.82:
      Successfully uninstalled nvidia-nvjitlink-cu12-12.5.82
  Attempting uninstall: nvidia-nccl-cu12
    Found existing installation: nvidia-nccl-cu12 2.23.4
    Uninstalling nvidia-nccl-cu12-2.23.4:
      Successfully uninstalled nvidia-nccl-cu12-2.23.4
  Attempting uninstall: nvidia-curand-cu12
    Found existing installation: nvidia-curand-cu12 10.3.6.82
    Uninstalling nvidia-curand-cu12-10.3.6.82:
      Successfully uninstalled nvidia-curand-cu12-10.3.6.82
  Attempting uninstall: nvidia-cuda-runtime-cu12
    Found existing installation: nvidia-cuda-runtime-cu12 12.5.82
    Uninstalling nvidia-cuda-runtime-cu12-12.5.82:
      Successfully uninstalled nvidia-cuda-runtime-cu12-12.5.82
  Attempting uninstall: nvidia-cuda-nvrtc-cu12
    Found existing installation: nvidia-cuda-nvrtc-cu12 12.5.82
    Uninstalling nvidia-cuda-nvrtc-cu12-12.5.82:
      Successfully uninstalled nvidia-cuda-nvrtc-cu12-12.5.82
  Attempting uninstall: nvidia-cuda-cupti-cu12
    Found existing installation: nvidia-cuda-cupti-cu12 12.5.82
    Uninstalling nvidia-cuda-cupti-cu12-12.5.82:
      Successfully uninstalled nvidia-cuda-cupti-cu12-12.5.82
  Attempting uninstall: nvidia-cublas-cu12
    Found existing installation: nvidia-cublas-cu12 12.5.3.2
    Uninstalling nvidia-cublas-cu12-12.5.3.2:
      Successfully uninstalled nvidia-cublas-cu12-12.5.3.2
  Attempting uninstall: nvidia-cusparse-cu12
    Found existing installation: nvidia-cusparse-cu12 12.5.1.3
    Uninstalling nvidia-cusparse-cu12-12.5.1.3:
      Successfully uninstalled nvidia-cusparse-cu12-12.5.1.3
  Attempting uninstall: nvidia-cufft-cu12
    Found existing installation: nvidia-cufft-cu12 11.2.3.61
    Uninstalling nvidia-cufft-cu12-11.2.3.61:
      Successfully uninstalled nvidia-cufft-cu12-11.2.3.61
  Attempting uninstall: nvidia-cudnn-cu12
    Found existing installation: nvidia-cudnn-cu12 9.3.0.75
    Uninstalling nvidia-cudnn-cu12-9.3.0.75:
      Successfully uninstalled nvidia-cudnn-cu12-9.3.0.75
  Attempting uninstall: nvidia-cusolver-cu12
    Found existing installation: nvidia-cusolver-cu12 11.6.3.83
    Uninstalling nvidia-cusolver-cu12-11.6.3.83:
      Successfully uninstalled nvidia-cusolver-cu12-11.6.3.83
  Attempting uninstall: torch
    Found existing installation: torch 2.6.0+cu124
    Uninstalling torch-2.6.0+cu124:
      Successfully uninstalled torch-2.6.0+cu124
  Attempting uninstall: datasets
    Found existing installation: datasets 4.0.0
    Uninstalling datasets-4.0.0:
      Successfully uninstalled datasets-4.0.0
  Attempting uninstall: torchvision
    Found existing installation: torchvision 0.21.0+cu124
    Uninstalling torchvision-0.21.0+cu124:
      Successfully uninstalled torchvision-0.21.0+cu124
ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.
fastai 2.7.19 requires torch<2.7,>=1.10, but you have torch 2.8.0 which is incompatible.
torchaudio 2.6.0+cu124 requires torch==2.6.0, but you have torch 2.8.0 which is incompatible.
Successfully installed bitsandbytes-0.47.0 cut_cross_entropy-25.1.1 datasets-3.6.0 msgspec-0.19.0 nvidia-cublas-cu12-12.8.4.1 nvidia-cuda-cupti-cu12-12.8.90 nvidia-cuda-nvrtc-cu12-12.8.93 nvidia-cuda-runtime-cu12-12.8.90 nvidia-cudnn-cu12-9.10.2.21 nvidia-cufft-cu12-11.3.3.83 nvidia-cufile-cu12-1.13.1.3 nvidia-curand-cu12-10.3.9.90 nvidia-cusolver-cu12-11.7.3.90 nvidia-cusparse-cu12-12.5.8.93 nvidia-cusparselt-cu12-0.7.1 nvidia-nccl-cu12-2.27.3 nvidia-nvjitlink-cu12-12.8.93 nvidia-nvtx-cu12-12.8.90 shtab-1.7.2 sympy-1.14.0 torch-2.8.0 torchvision-0.23.0 triton-3.4.0 trl-0.21.0 tyro-0.9.28 unsloth-2025.8.6 unsloth_zoo-2025.8.5 xformers-0.0.32.post2
Requirement already satisfied: bitsandbytes in /usr/local/lib/python3.11/dist-packages (0.47.0)
ERROR: Could not find a version that satisfies the requirement unsloth_zo (from versions: none)
ERROR: No matching distribution found for unsloth_zo

 

 

colab卸载torch 2.8.0，安装torch==2.6.0

在Google Colab中卸载并安装特定版本的PyTorch，可以通过以下步骤完成。首先，我们将卸载当前版本的PyTorch，然后安装所需版本的PyTorch。

步骤 1: 卸载当前版本的PyTorch

在Google Colab中，你可以使用!pip命令来运行pip命令，以卸载PyTorch。打开你的Colab笔记本，然后运行以下代码来卸载当前的PyTorch版本： 

!pip uninstall torch -y

这里，-y参数表示自动确认卸载，无需手动确认。

步骤 2: 安装指定版本的PyTorch

安装特定版本的PyTorch（例如2.6.0），你可以使用以下命令。根据你的需求（例如是否需要CUDA支持），你可以选择合适的命令。例如，如果你需要安装支持CUDA的PyTorch 2.6.0版本，可以使用：

  
!pip install torch==2.6.0+cu113 torchvision==0.13.0+cu113 torchaudio==0.12.0 --extra-index-url https://download.pytorch.org/whl/cu113

这里，cu113表示这是支持CUDA 11.3的版本。你需要根据你的GPU环境和CUDA版本选择合适的后缀。例如，如果你的环境是CUDA 10.2，则应该使用cu102作为后缀。例如：  

!pip install torch==2.6.0+cu102 torchvision==0.13.0+cu102 torchaudio==0.12.0 --extra-index-url https://download.pytorch.org/whl/cu102
 
https://www.cnblogs.com/emanlee/p/19041882
!pip install torch==2.6.0+cu124 torchvision==0.21.0+cu124 torchaudio==2.6.0 --index-url https://download.pytorch.org/whl/cu124
 
安装之后需要重启
 

检查安装

安装完成后，你可以通过以下Python代码来检查PyTorch的版本： 

import torch
print(torch.__version__)

这将会输出你刚刚安装的PyTorch版本号，确保它是2.6.0。

注意：

  • 确保你选择了与你的Colab环境兼容的CUDA版本。你可以在Colab的“Runtime”菜单下选择“Change runtime type”，然后在“Hardware accelerator”选项中选择合适的GPU类型（例如“GPU”或“Tesla T4”等），这有助于确定你需要哪个CUDA版本。

  • 如果你的环境中不需要CUDA（即使用CPU），你可以简单地运行：

      
!pip install torch==2.6.0 torchvision==0.13.0 torchaudio==0.12.0

 这将安装不包含CUDA支持的PyTorch版本。

 

检查colab中CUDA版本

!/opt/bin/nvidia-smi  和 torch.version.cuda 查得的cuda版本不一样。

图片

 


 
torch.version.cuda

    当你安装 PyTorch 时，它是针对特定版本的 CUDA Toolkit 进行编译的。如果你的 PyTorch 是编译时基于 CUDA Toolkit 11.7 版本，那么torch.version.cuda就会输出11.7，这个输出准确地反映了 PyTorch 所依赖的 CUDA Toolkit 版本。

nvidia - smi

    nvidia - smi输出的是显卡驱动的相关信息，其中包括该显卡驱动所支持的最高 CUDA 版本。例如，显卡驱动可能支持最高 CUDA 12.0 版本，但你在系统中实际安装并用于 PyTorch 的 CUDA Toolkit 可能是 11.7 版本。这是因为即使显卡驱动支持更高的 CUDA 版本，你依然可以选择安装和使用较低版本的 CUDA Toolkit 来配合你的深度学习框架和应用程序。所以nvidia - smi输出的 CUDA 版本相关信息可能大于（即支持更高版本）实际安装并用于特定框架（如 PyTorch）的 CUDA Toolkit 版本。

原文链接：https://blog.csdn.net/weixin_43501408/article/details/146635787
安装unsloth过程中，把pytorch 2.6卸载了，安装了 pytorch 2.8
pytorch 2.8和fastai 2.7.19 不兼容。
torchaudio 2.6.0+cu124 requires torch==2.6.0, but you have torch 2.8.0 which is incompatible.
fastai 2.7.19 requires torch<2.7,>=1.10, but you have torch 2.8.0 which is incompatible.
pytorch到pytorch2.6.0-cuda124，参考pytorch官网

 

!pip install torch==2.6.0+cu124 torchvision==0.21.0+cu124 torchaudio==2.6.0 --index-url https://download.pytorch.org/whl/cu124 【可行】

图片

 

安装unsloth时，fastai 2.7.19 requires torch<2.7,>=1.10, but you have torch 2.8.0 which is incompatible. 

 

https://github.com/unslothai/unsloth   

pip install --upgrade pip
pip install "unsloth[cu121-torch240] @ git+https://github.com/unslothai/unsloth.git"

pip install "unsloth[cu124-torch260] @ git+https://github.com/unslothai/unsloth.git"  【可行】

 

xformers巨坑
安装torch一切岁月静好，但用同样的命令安装 xformers时，你会发现，没错，你的torch被重装了，我这里被重装为2.2.2，这也是前期说不用查询torchvision和torch的版本对应关系去下载的原因，因为白费功夫，torch重装后会报错torchvision不对应，因此，不如不装，先把 xformers搞定。血淋淋的torch被重装和报错torchvision不对应的现场图如下。没错，查询好版本对应也不好使。没事，先把xformers装上再解决别的问题。                       
原文链接：https://blog.csdn.net/weixin_47177911/article/details/145937705

 

💾 Install Unsloth

You can also see our documentation for more detailed installation and updating instructions here.

Pip Installation

Install with pip (recommended) for Linux devices:

pip install unsloth
 

To update Unsloth:

pip install --upgrade --force-reinstall --no-cache-dir unsloth unsloth_zoo
 

See here for advanced pip install instructions.

Windows Installation

Warning

Python 3.13 does not support Unsloth. Use 3.12, 3.11 or 3.10

  1. Install NVIDIA Video Driver: You should install the latest version of your GPUs driver. Download drivers here: NVIDIA GPU Drive.

  2. Install Visual Studio C++: You will need Visual Studio, with C++ installed. By default, C++ is not installed with Visual Studio, so make sure you select all of the C++ options. Also select options for Windows 10/11 SDK. For detailed instructions with options, see here.

  3. Install CUDA Toolkit: Follow the instructions to install CUDA Toolkit.

  4. Install PyTorch: You will need the correct version of PyTorch that is compatible with your CUDA drivers, so make sure to select them carefully. Install PyTorch.

  5. Install Unsloth:

pip install unsloth
 

Notes

To run Unsloth directly on Windows:

  • Install Triton from this Windows fork and follow the instructions here (be aware that the Windows fork requires PyTorch >= 2.4 and CUDA 12)
  • In the SFTConfig, set dataset_num_proc=1 to avoid a crashing issue:
SFTConfig(
    dataset_num_proc=1,
    ...
)
 

Advanced/Troubleshooting

For advanced installation instructions or if you see weird errors during installations:

  1. Install torch and triton. Go to https://pytorch.org to install it. For example pip install torch torchvision torchaudio triton
  2. Confirm if CUDA is installed correctly. Try nvcc. If that fails, you need to install cudatoolkit or CUDA drivers.
  3. Install xformers manually. You can try installing vllm and seeing if vllm succeeds. Check if xformers succeeded with python -m xformers.info Go to https://github.com/facebookresearch/xformers. Another option is to install flash-attn for Ampere GPUs.
  4. Double check that your versions of Python, CUDA, CUDNN, torch, triton, and xformers are compatible with one another. The PyTorch Compatibility Matrix may be useful.
  5. Finally, install bitsandbytes and check it with python -m bitsandbytes

Conda Installation (Optional)

⚠️Only use Conda if you have it. If not, use Pip. Select either pytorch-cuda=11.8,12.1 for CUDA 11.8 or CUDA 12.1. We support python=3.10,3.11,3.12.

conda create --name unsloth_env \
    python=3.11 \
    pytorch-cuda=12.1 \
    pytorch cudatoolkit xformers -c pytorch -c nvidia -c xformers \
    -y
conda activate unsloth_env

pip install unsloth
 

If you're looking to install Conda in a Linux environment, read here, or run the below 🔽

 

Advanced Pip Installation

⚠️Do **NOT** use this if you have Conda. Pip is a bit more complex since there are dependency issues. The pip command is different for torch 2.2,2.3,2.4,2.5 and CUDA versions.

For other torch versions, we support torch211, torch212, torch220, torch230, torch240 and for CUDA versions, we support cu118 and cu121 and cu124. For Ampere devices (A100, H100, RTX3090) and above, use cu118-ampere or cu121-ampere or cu124-ampere.

For example, if you have torch 2.4 and CUDA 12.1, use:

pip install --upgrade pip
pip install "unsloth[cu121-torch240] @ git+https://github.com/unslothai/unsloth.git"
 

Another example, if you have torch 2.5 and CUDA 12.4, use:

pip install --upgrade pip
pip install "unsloth[cu124-torch250] @ git+https://github.com/unslothai/unsloth.git"
 

And other examples:

pip install "unsloth[cu121-ampere-torch240] @ git+https://github.com/unslothai/unsloth.git"
pip install "unsloth[cu118-ampere-torch240] @ git+https://github.com/unslothai/unsloth.git"
pip install "unsloth[cu121-torch240] @ git+https://github.com/unslothai/unsloth.git"
pip install "unsloth[cu118-torch240] @ git+https://github.com/unslothai/unsloth.git"

pip install "unsloth[cu121-torch230] @ git+https://github.com/unslothai/unsloth.git"
pip install "unsloth[cu121-ampere-torch230] @ git+https://github.com/unslothai/unsloth.git"

pip install "unsloth[cu121-torch250] @ git+https://github.com/unslothai/unsloth.git"
pip install "unsloth[cu124-ampere-torch250] @ git+https://github.com/unslothai/unsloth.git"
 

Or, run the below in a terminal to get the optimal pip installation command:

wget -qO- https://raw.githubusercontent.com/unslothai/unsloth/main/unsloth/_auto_install.py | python -
 

Or, run the below manually in a Python REPL:

try: import torch
except: raise ImportError('Install torch via `pip install torch`')
from packaging.version import Version as V
v = V(torch.__version__)
cuda = str(torch.version.cuda)
is_ampere = torch.cuda.get_device_capability()[0] >= 8
if cuda != "12.1" and cuda != "11.8" and cuda != "12.4": raise RuntimeError(f"CUDA = {cuda} not supported!")
if   v <= V('2.1.0'): raise RuntimeError(f"Torch = {v} too old!")
elif v <= V('2.1.1'): x = 'cu{}{}-torch211'
elif v <= V('2.1.2'): x = 'cu{}{}-torch212'
elif v  < V('2.3.0'): x = 'cu{}{}-torch220'
elif v  < V('2.4.0'): x = 'cu{}{}-torch230'
elif v  < V('2.5.0'): x = 'cu{}{}-torch240'
elif v  < V('2.6.0'): x = 'cu{}{}-torch250'
else: raise RuntimeError(f"Torch = {v} too new!")
x = x.format(cuda.replace(".", ""), "-ampere" if is_ampere else "")
print(f'pip install --upgrade pip && pip install "unsloth[{x}] @ git+https://github.com/unslothai/unsloth.git"')
 


 

posted @ 2025-08-16 11:00  emanlee  阅读(128)  评论(0)    收藏  举报