TabPFN: Add ignore_pretraining_limits to classifiers and regressors

[anuprulez]

The parameter ignore_pretraining_limits=True removes the limit to use only 1000 samples for training.

PriorLabs/TabPFN#169
https://huggingface.co/Prior-Labs/TabPFN-v2-reg/discussions/2

Currently, the tool fails when training size > 1000 samples:

Traceback (most recent call last):
  File "/opt/galaxy/shed_tools/toolshed.g2.bx.psu.edu/repos/bgruening/tabpfn/ed78e1448387/tabpfn/main.py", line 167, in <module>
    train_evaluate(args)
  File "/opt/galaxy/shed_tools/toolshed.g2.bx.psu.edu/repos/bgruening/tabpfn/ed78e1448387/tabpfn/main.py", line 98, in train_evaluate
    classifier.fit(tr_features, tr_labels)
  File "/usr/local/lib/python3.12/contextlib.py", line 81, in inner
    return func(*args, **kwds)
           ^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.12/site-packages/tabpfn_common_utils/telemetry/core/decorators.py", line 321, in wrapper
    return _safe_call_with_telemetry(
           ^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.12/site-packages/tabpfn_common_utils/telemetry/core/decorators.py", line 365, in _safe_call_with_telemetry
    result = func(*args, **kwargs)
             ^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.12/site-packages/tabpfn/classifier.py", line 758, in fit
    ensemble_configs, X, y = self._initialize_dataset_preprocessing(
                             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.12/site-packages/tabpfn/classifier.py", line 642, in _initialize_dataset_preprocessing
    X, y, feature_names, n_features, original_y_name = ensure_compatible_fit_inputs(
                                                       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.12/site-packages/tabpfn/validation.py", line 75, in ensure_compatible_fit_inputs
    validate_dataset_size(
  File "/usr/local/lib/python3.12/site-packages/tabpfn/validation.py", line 137, in validate_dataset_size
    _validate_num_samples_for_cpu(
  File "/usr/local/lib/python3.12/site-packages/tabpfn/validation.py", line 274, in _validate_num_samples_for_cpu
    raise RuntimeError(
RuntimeError: Running on CPU with more than 1000 samples is not allowed by default due to slow performance.
To override this behavior, set the environment variable TABPFN_ALLOW_CPU_LARGE_DATASET=1 or set ignore_pretraining_limits=True.
Alternatively, consider using a GPU or the tabpfn-client API: https://github.com/PriorLabs/tabpfn-client

[bgruening]

Are there any disadvantages here? Should that be a user option? Or a option that we toggle based on input size?

[anuprulez]

Are there any disadvantages here?

It will take more time to run if > 1000 examples. So, I selected GPU option to run but it produced the error RuntimeError: Running on CPU with more than 1000 samples is not allowed by default. Not sure why it still considered the tool was running on CPUs, it executed on a GPU node (c64m384g8-n3801.bi.privat)

Should that be a user option? Or a option that we toggle based on input size?

Yes, we can add it directly in the tool based on the user's training data (toggle in the script).

if X.shape[0] <= 1000
    TabPFNClassifier()
else:
    TabPFNClassifier(ignore_pretraining_limits=True)

[anuprulez]

Can someone add skip-version-check label to allow tests to pass? I don't have necessary permissions. Thanks a lot!

Linting repository /home/runner/work/galaxytools/galaxytools/tools/tabpfn
.. INFO: Included files all found.
.. INFO: No tool_dependencies.xml, skipping.
.. INFO: No tool_dependencies.xml, skipping.
.. INFO: No tool_dependencies.xml, skipping.
.. INFO: No repository_dependencies.xml, skipping.
.. INFO: .shed.yml found and appears to be valid YAML.
.. INFO: No README found, skipping.
.. INFO: No tool_dependencies.xml, skipping.
+Linting tool /home/runner/work/galaxytools/galaxytools/tools/tabpfn/tabpfn.xml
.. CHECK (TestsNoValid): 8 test(s) found.
.. INFO (OutputsNumber): 2 outputs found.
.. INFO (InputsNum): Found 9 input parameters.
.. CHECK (HelpPresent): Tool contains help section.
.. CHECK (HelpValidRST): Help contains valid reStructuredText.
.. CHECK (ToolIDValid): Tool defines an id [tabpfn].
.. CHECK (ToolNameValid): Tool defines a name [Tabular data prediction using TabPFN].
.. CHECK (ToolProfileValid): Tool specifies profile version [24.2].
.. CHECK (ToolVersionValid): Tool defines a version [7.0.0+galaxy1].
.. INFO (CommandInfo): Tool contains a command.
.. CHECK (CitationsFound): Found 1 citations.
.. INFO: URL OK https://github.com/PriorLabs/TabPFN?tab=License-1-ov-file
.. INFO: URL OK https://huggingface.co/Prior-Labs/tabpfn_2_5/blob/main/LICENSE
.. ERROR (ShedVersion): tabpfn: version 7.0.0+galaxy1 is less or equal than version of the latest installable revision 7.0.0+galaxy1
.. INFO: Found all shed metadata fields required for automated repository creation and/or updates.