← Back to all news
arXiv Computation and Language By Yawei Li, Jiazheng Li, David R\"ugamer, Bernd Bischl, Yulan He, Mina Rezaei

S-EDL: Eliciting Self-Evidence from Sequence Likelihoods for Semantic Calibration of LLMs

Read the original on arXiv Computation and Language →

S-EDL is a new evidential method that calibrates large language models for open‑ended generation by using the model’s own sequence likelihoods to build a differentiable Dirichlet prior over prompt‑specific semantic classes. This approach optimizes an evidential loss that provides a calibration‑aware training signal without relying on fixed multiple‑choice options. Experiments on four open‑ended benchmarks and three models show that S-EDL reduces semantic calibration error and improves accuracy, offering a framework for trustworthy LLM deployment in complex applications.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Computation and Language.

Related stories

arXiv Computation and Language
Sep 23

Calibration as a First-Class Criterion in LLM Evaluation

The paper argues that calibration—how well a language model’s confidence aligns with its actual correctness—should be a standard evaluation metric for large language models (LLMs). It notes that while calibration metrics exist, they are rarely applied outside specialized NLP subfields, leading to unverified confidence scores in new models, datasets, and benchmarks. The authors highlight the risks of miscalibration both at deployment (overconfident errors causing harm) and in research workflows (affecting LLM-as-a-judge, synthetic data generation, and active learning). They call for every NLP subfield to pair its primary performance metric with a calibration score, treating calibration as an essential property of every model.

By Mario Sanz-Guerrero, Katharina von der Wense
llmsnlpbenchmarkssafety
More like this →
arXiv Machine Learning
Sep 21

Semantic Calibration Prevails Where Token Confidence Fails: Benchmarking Long-Form Scientific QA

The paper presents the first large‑scale benchmark for uncertainty quantification (UQ) calibration in long‑form scientific question answering, evaluating four UQ methods on 685,000 responses from up to 20 large language models across seven datasets. It shows that instruction tuning leads to token‑level probability polarization, undermining token‑level uncertainty signals, while reasoning model families differ in how they handle this effect. Only semantic consistency—consistency of the final answer—provides well‑calibrated outputs, demonstrating that semantic calibration remains robust in multi‑step, dependency‑rich reasoning.

By Philip M\"uller, Nicholas Popovi\v{c}, Michael F\"arber, Peter Steinbach
llmsnlpfine-tuningbenchmarks
More like this →
arXiv AI
Aug 11

Embedding Trust: Semantic Isotropy Predicts Nonfactuality in Long-Form Text Generation

arXiv:2510. 21891v2 Announce Type: replace-cross Abstract: To deploy large language models (LLMs) in high-stakes application domains that require substantively accurate responses to open-ended prompts, we need reliable, computationally inexpensive methods that assess the trustworthiness of long-form responses generated by LLMs.

By Dhrupad Bhardwaj, Julia Kempe, Tim G. J. Rudner
llmsragfine-tuning
More like this →
arXiv Machine Learning
Jun 5

Can LLMs Write Correct TLA+ Specifications? Evaluating Natural-Language-to-TLA+ Generation

arXiv:2606. 05792v1 Announce Type: cross Abstract: TLA+ has supported industrial verification at companies such as Amazon and Microsoft, yet writing correct TLA+ specifications from natural language still requires time and expertise, which limits adoption.

By Arslan Bisharat, Brian Ortiz, Eric Spencer, Khushboo Bhadauria, TaiNing Wang, George K. Thiruvathukal, Konstantin Laufer, Mohammed Abuhamad
llmssafety
More like this →