hermanda/robeczech-propaganda

Update model card for v2: calibration table, new thresholds, training data source

Update to v2: retrained on human-reviewed Argilla labels (comment-only, full finetune, 2x oversample)

Update to v2: retrained on human-reviewed Argilla labels (comment-only, full finetune, 2x oversample)

Update to v2: retrained on human-reviewed Argilla labels (comment-only, full finetune, 2x oversample)

Update to v2: retrained on human-reviewed Argilla labels (comment-only, full finetune, 2x oversample)

v4: 27k training data, all layers, class_weight=3.0, lr=1e-5, early_stop on loss

Upload tokenizer

v3: all layers, lr=1e-5, class_weight=3.0, early_stop on loss, F1=0.505

initial commit