Hi
Has anyone had issues installing pytorch using conda? We have an on-premises Anacoda repo (air gapped /w no Internet) and I am unable to install pytorch using ‘conda install pytorch’ in a clean environment.
conda create -n pytorch
conda activate pytorch
conda clean --all
conda install pytorch (or pytorch-cpu)
I get this error
>>>
conda.CondaMultiError: Error with archive C:\conda_pkgs\pytorch-2.10.0-cpu_mkl_py314hf6a6a2c_103.conda. You probably n:
failed with error: [Errno 2] No such file or directory: ‘C:\conda_pkgs\pytorch-2.10.0-cpu_mkl_py314hf6a6a2c_103\info’
>>>
>>>
Downloading and Extracting Packages:
InvalidArchiveError("Error with archive C:\conda_pkgs\pytorch-2.10.0-cpu_mkl_py314hf6a6a2c_103.conda. You probably need to delete and re-download or re-create this file. Message was:\n\nfailed with error: [Errno 2] )
>>>
I can install all package dependencies except for pytorch itself. Other users have installed it so I am not sure if the .conda file itself might be corrupt in the repo. Installing other versions of pytorch < 2.10 results in the same error.
Reinstalling conda did not fix the problem.
The version in the Anaconda repo is pytorch(-cpu) v2.10.0 and was last updated May 6, 2026
I do not have access to the public conda-forge repo, only the curated on-premises repo.
Pytorch support indicated that they no longer publishing the pytorch binaries on conda: [Announcement] Deprecating PyTorch’s official Anaconda channel · Issue #138506 · pytorch/pytorch · GitHub
With no Internet access the Anaconda repo default channels are my only option.
Any suggestions?
Take care,
Shayne
Hi @Shayne ,
A couple of things to clarify and some next steps based on my current understanding of what is happening.
On PyTorch availability: The GitHub issue you linked refers to PyTorch deprecating their own Anaconda channel (pytorch). That’s separate from the Anaconda defaults channel, where we independently build and publish PyTorch. Defaults remains a supported path — you’re not on a dead-end here. We just released PyTorch 2.11 this morning 
.
On the extraction error: The InvalidArchiveError is an archive integrity issue, and the error message is actually pointing you in the right direction: You probably need to delete and re-download or re-create this file. The catch in an air-gapped environment is that conda clean --all and re-downloading on the client side won’t help — your client is just re-pulling the same potentially incomplete archive from the on-prem mirror. The fix needs to happen one layer up: the on-prem repo itself needs to re-mirror/re-sync the PyTorch package from the Anaconda source. PyTorch packages are large, and partial transfers during the mirror sync are the most common cause of this pattern.
I’d recommend opening a support ticket at https://support.anaconda.com — where your team can get dedicated support.
Hello @Shayne,
Regarding InvalidArchiveError , this is a windows related issue (LongPaths) not Anaconda.
There are two ways to resolve this. I will include the 1st method with the use of the Group Policy Editor below and a Support article that will assist you with more information about creating or editing the LongPathsEnabled file in the Registry Editor.
Please note that both methods will require admin access to perform.
Method 1:
Using Group Policy Editor (For Windows 10 Pro, Enterprise, and Education)
Press Windows + R to open the Run dialog.
Type gpedit.msc and press Enter to open the Local Group Policy Editor.
Navigate to the following path: Local Computer Policy > Computer Configuration > Administrative Templates > System > Filesystem
Find and double-click on “Enable Win32 long paths”.
In the window that appears, select Enabled.
Click OK and then Apply.
Reboot your computer
After enabling this, Windows will allow applications to handle long file paths.
Method 2:
The second method is included in the article link here. This has a step by step process to see if your system has LongPathsEnabled. This method works for Windows 10 and newer systems.
Once LongPathEnabled setting has been enabled, you will need to reboot your computer. After your computer has been rebooted, Please re-run the conda commands that prompted the error. It should be resolved.
1 Like
This error usually points to a corrupted or incomplete .conda package inside your internal Anaconda repository, especially since the extraction fails before installation begins. The missing info directory strongly suggests the PyTorch archive was not mirrored or cached correctly on the air-gapped repo. Try deleting the affected package from C:\conda_pkgs\ and ask your repo admin to re-sync or re-upload the PyTorch build. Since older versions fail too, the local channel metadata may also be broken. Similar to how deltaexxecutor provides updated and reliable resource management for users, your best fix here is validating the integrity of the repo packages rather than reinstalling Conda itself.
Thanks, this was the fix I needed!
1 Like