environment setup has invalid section in yaml and dependency conflicts on Ubuntu 22.04 with CUDA GPU

[rbavery]

👋 I'm poking around the yaml to see what the fix is for my machine, not sure if others are running into this. I'm using the latest version of mamba.

# rave at rave-desktop in ~/clay-model on git:main ● [15:10:19]
→ mamba env create --file environment.yml

EnvironmentSectionNotValid: The following section on '/home/rave/clay-model/environment.yml' is invalid and will be ignored:
 - platforms

Channels:
 - conda-forge
Platform: linux-64
Collecting package metadata (repodata.json): done
Solving environment: failed
Channels:
 - conda-forge
Platform: linux-64
Collecting package metadata (repodata.json): done
Solving environment: failed

LibMambaUnsatisfiableError: Encountered problems while solving:
  - package torchgeo-0.5.2-pyhd8ed1ab_0 requires segmentation-models-pytorch >=0.2, but none of the providers can be installed
  - package pylance-0.14.1-py311h6a0c370_0 requires pyarrow >=12,<15.0.1, but none of the providers can be installed

Could not solve for environment specs
The following packages are incompatible
├─ lancedb ~=0.10.2  is installable with the potential options
│  ├─ lancedb 0.10.2 would require
│  │  └─ pylance 0.14.1  with the potential options
│  │     ├─ pylance 0.14.1 would require
│  │     │  └─ pyarrow >=12,<15.0.1  with the potential options
│  │     │     ├─ pyarrow [12.0.0|12.0.1|...|16.1.0] would require
│  │     │     │  └─ python_abi 3.10.* *_cp310, which can be installed;
│  │     │     ├─ pyarrow [12.0.0|12.0.1|...|15.0.0], which can be installed;
│  │     │     ├─ pyarrow [12.0.0|12.0.1|...|16.1.0] would require
│  │     │     │  └─ python_abi 3.8.* *_cp38, which can be installed;
│  │     │     ├─ pyarrow [12.0.0|12.0.1|...|16.1.0] would require
│  │     │     │  └─ python_abi 3.9.* *_cp39, which can be installed;
│  │     │     └─ pyarrow [13.0.0|14.0.0|...|16.1.0] would require
│  │     │        └─ python_abi 3.12.* *_cp312, which can be installed;
│  │     ├─ pylance 0.14.1 would require
│  │     │  └─ python_abi 3.10.* *_cp310, which can be installed;
│  │     ├─ pylance 0.14.1 would require
│  │     │  └─ python_abi 3.12.* *_cp312, which can be installed;
│  │     └─ pylance 0.14.1 would require
│  │        └─ python_abi 3.9.* *_cp39, which can be installed;
│  ├─ lancedb 0.10.2 would require
│  │  └─ python_abi 3.10.* *_cp310, which can be installed;
│  ├─ lancedb 0.10.2 would require
│  │  └─ python_abi 3.12.* *_cp312, which can be installed;
│  └─ lancedb 0.10.2 would require
│     └─ python_abi 3.9.* *_cp39, which can be installed;
├─ pyarrow ~=16.1.0  is installable with the potential options
│  ├─ pyarrow [12.0.0|12.0.1|...|16.1.0], which can be installed (as previously explained);
│  ├─ pyarrow [12.0.0|12.0.1|...|16.1.0], which can be installed (as previously explained);
│  ├─ pyarrow [12.0.0|12.0.1|...|16.1.0], which can be installed (as previously explained);
│  ├─ pyarrow [13.0.0|14.0.0|...|16.1.0], which can be installed (as previously explained);
│  └─ pyarrow 16.1.0 conflicts with any installable versions previously reported;
├─ python ~=3.11.0  is not installable because it requires
│  └─ python_abi 3.11.* *_cp311, which conflicts with any installable versions previously reported;
├─ timm ~=0.9.16  is requested and can be installed;
└─ torchgeo ~=0.5.2  is not installable because it requires
   └─ segmentation-models-pytorch >=0.2  but there are no viable options
      ├─ segmentation-models-pytorch [0.2.0|0.2.1|0.3.0|0.3.1] would require
      │  └─ timm 0.4.12 , which conflicts with any installable versions previously reported;
      ├─ segmentation-models-pytorch 0.3.2 would require
      │  └─ timm 0.6.12 , which conflicts with any installable versions previously reported;
      └─ segmentation-models-pytorch 0.3.3 would require
         └─ timm 0.9.2 , which conflicts with any installable versions previously reported.

[rbavery]

I can install from the lockfile successfully following these instructions, might be an upstream issue with mamba or how a dependency specifies it's own dependencies. https://clay-foundation.github.io/model/getting-started/installation.html#advanced

[weiji14]

You're hitting into 2 separate dependency solving issues:

LibMambaUnsatisfiableError: Encountered problems while solving:
  - package torchgeo-0.5.2-pyhd8ed1ab_0 requires segmentation-models-pytorch >=0.2, but none of the providers can be installed
  - package pylance-0.14.1-py311h6a0c370_0 requires pyarrow >=12,<15.0.1, but none of the providers can be installed

The first one is because torchgeo depends on smp which has a hard pin on timm=0.9.2 (https://github.com/conda-forge/segmentation-models-pytorch-feedstock/pull/22#discussion_r1511659865). Second one is lancedb=0.10.2 depends on pylance=0.14.1 which has a upper pin on pyarrow<15.0.1.

I've opened a PR to refresh the lockfile at https://github.com/Clay-foundation/model/pull/310. Could you test installing from that branch and run through a few examples?

[jonm3D]

The new lockfile worked for me (macOS).