Skip to content

Results with multiversion coudn't be loaded #4830

Description

@Samoed

Describe the bug

2026-06-18 14:37:44,123 - INFO - Starting to create PR results comment...
2026-06-18 14:37:44,123 - INFO - Using reference models: intfloat/multilingual-e5-large, google/gemini-embedding-001
Traceback (most recent call last):
  File "/home/runner/work/results/results/scripts/create_pr_results_comment.py", line 350, in <module>
    main(args.reference_models, args.output)
  File "/home/runner/work/results/results/scripts/create_pr_results_comment.py", line 342, in main
    markdown = generate_markdown_content(model_tasks, reference_models)
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/runner/work/results/results/scripts/create_pr_results_comment.py", line 295, in generate_markdown_content
    df, high_model_performance_tasks, all_training_datasets = create_comparison_table(
                                                              ^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/runner/work/results/results/scripts/create_pr_results_comment.py", line 117, in create_comparison_table
    results = cache.load_results(models=models, tasks=tasks, validate_and_filter=True)
              ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/runner/work/results/results/.venv/lib/python3.12/site-packages/mteb/cache/result_cache.py", line 1389, in load_results
    task_result = TaskResult.from_disk(path)
                  ^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/runner/work/results/results/.venv/lib/python3.12/site-packages/mteb/results/task_result.py", line 410, in from_disk
    and Version(data["mteb_version"]) < Version("1.11.0")
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/runner/work/results/results/.venv/lib/python3.12/site-packages/packaging/version.py", line 438, in __init__
    raise InvalidVersion(f"Invalid version: {version!r}")
packaging.version.InvalidVersion: Invalid version: '2.12.16-2.15.4'

To reproduce

https://github.com/embeddings-benchmark/results/actions/runs/27767067343/job/82156582774?pr=574

Additional information

No response

Are you interested to contribute a fix for this bug?

Yes

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No fields configured for Bug.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions