الملخص
Pre-generation success probes estimate response correctness from a language model's hidden activations before decoding, enabling cost-aware routing. While prior work has demonstrated their utility primarily on English inputs, we study their reliability across languages along three dimensions: (1) whether they preserve the ranking of likely successes and failures (DISCRIMINATION); (2) whether they retain probabilities that match observed success frequencies (CALIBRATION); and (3) whether they produce scores comparable enough across candidate models for cost-aware multilingual routing (UTILITY). Using 3,000 MATH problems in 10 languages and 8 open-weight model configurations, we compare cross-lingual transfer from English-trained probes and equal-budget pooled multilingual probes. English-trained probes retain useful cross-lingual discrimination but become less well calibrated after transfer. Pooled multilingual supervision improves both properties and yields more reliable estimates of success. In routing experiments, the pooled router achieves a 0.7% higher test success rate while reducing modeled cost by 13.0% relative to always selecting the model with the highest average success. These results show that multilingual routing requires success estimates that remain well calibrated and comparable across languages and models.
الكلمات المفتاحية
الموضوع
بيانات النشر
- المجلة
- غير متاح
- وصول مفتوح
- وصول مفتوح أخضر
اقتبس هذه المقالة
APA 7
Paganelli, A., Civelli, S., Bernardelle, P., & Demartini, G. (2026). Cross-lingual Calibration of Pre-Generation Success Probes for Multilingual LLM Routing. https://omanscience.com/ar/articles/cross-lingual-calibration-of-pre-generation-success-probes-for-multilingual-llm-routing
MLA 9
Paganelli, Andrea, et al. "Cross-lingual Calibration of Pre-Generation Success Probes for Multilingual LLM Routing." https://omanscience.com/ar/articles/cross-lingual-calibration-of-pre-generation-success-probes-for-multilingual-llm-routing.
شيكاغو (المؤلف–التاريخ)
Paganelli, Andrea, Stefano Civelli, Pietro Bernardelle, and Gianluca Demartini. 2026. "Cross-lingual Calibration of Pre-Generation Success Probes for Multilingual LLM Routing." https://omanscience.com/ar/articles/cross-lingual-calibration-of-pre-generation-success-probes-for-multilingual-llm-routing.
هارفارد
Paganelli, A., Civelli, S., Bernardelle, P. and Demartini, G. (2026) 'Cross-lingual Calibration of Pre-Generation Success Probes for Multilingual LLM Routing', Available at: https://omanscience.com/ar/articles/cross-lingual-calibration-of-pre-generation-success-probes-for-multilingual-llm-routing.
فانكوفر
Paganelli A, Civelli S, Bernardelle P, Demartini G. Cross-lingual Calibration of Pre-Generation Success Probes for Multilingual LLM Routing. https://omanscience.com/ar/articles/cross-lingual-calibration-of-pre-generation-success-probes-for-multilingual-llm-routing
IEEE
A. Paganelli, S. Civelli, P. Bernardelle, and G. Demartini, "Cross-lingual Calibration of Pre-Generation Success Probes for Multilingual LLM Routing," https://omanscience.com/ar/articles/cross-lingual-calibration-of-pre-generation-success-probes-for-multilingual-llm-routing.