| Previous changeset 9:ed78e1448387 (2026-04-20) |
|
Commit message:
planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/tabpfn commit 5f1f7b83ced6c25d0024de3fbcc63f3f9e25373f |
|
modified:
main.py |
| b |
| diff -r ed78e1448387 -r f0c7f0bad621 main.py --- a/main.py Mon Apr 20 08:08:59 2026 +0000 +++ b/main.py Wed Apr 22 21:52:07 2026 +0000 |
| [ |
| @@ -84,6 +84,8 @@ """ Train TabPFN and predict """ + MAX_IGNORE_PRETRAINING_LIMITS_SAMPLES = 1000 + SEED = 42 # prepare train data tr_features, tr_labels = separate_features_labels(args["train_data"], args["train_header"]) # prepare test data @@ -94,7 +96,10 @@ te_labels = [] s_time = time.time() if args["selected_task"] == "Classification": - classifier = TabPFNClassifier(random_state=42, model_path=args["model_path"]) + if tr_features.shape[0] <= MAX_IGNORE_PRETRAINING_LIMITS_SAMPLES: + classifier = TabPFNClassifier(random_state=SEED, model_path=args["model_path"]) + else: + classifier = TabPFNClassifier(random_state=SEED, model_path=args["model_path"], ignore_pretraining_limits=True) classifier.fit(tr_features, tr_labels) y_eval = classifier.predict(te_features) pred_probas_test = classifier.predict_proba(te_features) @@ -105,7 +110,10 @@ "output_predicted_data", sep="\t", index=None ) else: - regressor = TabPFNRegressor(random_state=42, model_path=args["model_path"]) + if tr_features.shape[0] <= MAX_IGNORE_PRETRAINING_LIMITS_SAMPLES: + regressor = TabPFNRegressor(random_state=SEED, model_path=args["model_path"]) + else: + regressor = TabPFNRegressor(random_state=SEED, model_path=args["model_path"], ignore_pretraining_limits=True) regressor.fit(tr_features, tr_labels) y_eval = regressor.predict(te_features) if len(te_labels) > 0: |