ZeroHour
Hugging Face daily paperspublished ()ingested Nagham Omar, Maya Rozenshtein, Evgeny Mishlyakov

Train Smarter, Not Harder: Switching Signal-Guided Training in Active Learning

infoAI researchimportance 20
AI summary · glm-5.3-flash

HybridAL is an active-learning training schedule that switches from retraining to fine-tuning on stabilization signals, saving up to 49% time.

Researchers find that choosing between retraining from scratch and fine-tuning is an exploitable decision variable in active learning: retraining helps in early rounds while fine-tuning is safer once the model trajectory stabilizes. HybridAL monitors an online stabilization signal using spectral exponent change and accuracy change, switching from retraining to fine-tuning after sustained stabilization. Across three encoder backbones and six text-classification tasks with five seeds each, HybridAL keeps endpoint macro-F1 non-inferior within a 0.010 margin, saves up to 49% of retraining time, and improves the time-calibration trade-off measured by negative log-likelihood.

  • Shows retraining versus fine-tuning choice is exploitable in active learning
  • HybridAL switches from retraining to fine-tuning after sustained stabilization
  • Uses spectral exponent change and accuracy change as switching signals
  • Saves up to 49% of retraining time with non-inferior macro-F1
  • Outperforms fixed-round switching on NLL at moderate extra cost
ProductsHybridAL
Full article173 words · extracted from huggingface.co · click to collapse

Training strategy, namely whether to retrain from scratch or fine-tune from the previous checkpoint, is an overlooked decision variable in active learning. We show that this choice has exploitable structure: retraining is most useful in early rounds, when each batch can substantially reshape the labeled distribution, while fine-tuning becomes safer once the model trajectory stabilizes. We propose HybridAL, an adaptive training schedule that monitors an online stabilization signal and switches from retraining to fine-tuning after sustained stabilization. Two complementary signals, spectral exponent change Δα (weight-based) and accuracy change ΔAcc (validation-based), span different points on the time-calibration trade-off. Across three encoder backbones and six text-classification tasks (five seeds each), HybridAL keeps endpoint macro-F1 non-inferior to retraining and fine-tuning at a 0.010 margin, saves up to 49% of retraining time, and recovers a substantial fraction of retraining's calibration advantage as measured by negative log-likelihood (NLL). Compared with schedules that switch at a pre-committed round, HybridAL obtains lower NLL at moderate additional cost, showing that trajectory-dependent switching provides a stronger time-calibration trade-off than fixed early switching.

Text extracted automatically; images, tables and formatting may be missing. Original: https://huggingface.co/papers/2609.06806