Saved in:
Bibliographic Details
Main Authors: Gajjar, Pranshav, Ojo, Emmanuel, Shah, Vijay K
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2605.09929
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911670249455616
author Gajjar, Pranshav
Ojo, Emmanuel
Shah, Vijay K
author_facet Gajjar, Pranshav
Ojo, Emmanuel
Shah, Vijay K
contents Deploying large language models in telecommunications requires more than task accuracy. In realistic workflows, a model may inherit partially completed reasoning from a prior step, an upstream agent, or its own earlier generation, and must continue that reasoning even when it is already going wrong. We introduce TeleResilienceBench, a benchmark that quantifies this capability, which we term reasoning resilience, across seven telecom sub-domains drawn from the GSMA Open-Telco LLM suite. Instances are constructed by collecting failures from a weak generator model, truncating the flawed reasoning trace at its midpoint, and asking a target model to continue and correct it. We propose the Correct Flip Rate (CFR) as a direct measure of successful recovery and evaluate eight models spanning the Qwen3.5, Gemma4, and Nemotron-3 families. Our results show that even the strongest model achieves a macro-average CFR of only 29.1%, and scale does not reliably improve resilience within families. Nemotron-3-nano 4b outperforms all Qwen3.5 variants including the 27b model and leads the auxiliary TeleMath numerical evaluation at 23.4% CR%, offering the best resilience-to-cost ratio in the set. A difficulty-stratified analysis further reveals that existing telecom benchmark difficulty labels reflect factual specificity rather than reasoning depth, suggesting that current evaluations measure knowledge coverage more than reasoning ability.
format Preprint
id arxiv_https___arxiv_org_abs_2605_09929
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle TeleResilienceBench: Quantifying Resilience for LLM Reasoning in Telecommunications
Gajjar, Pranshav
Ojo, Emmanuel
Shah, Vijay K
Machine Learning
Software Engineering
Deploying large language models in telecommunications requires more than task accuracy. In realistic workflows, a model may inherit partially completed reasoning from a prior step, an upstream agent, or its own earlier generation, and must continue that reasoning even when it is already going wrong. We introduce TeleResilienceBench, a benchmark that quantifies this capability, which we term reasoning resilience, across seven telecom sub-domains drawn from the GSMA Open-Telco LLM suite. Instances are constructed by collecting failures from a weak generator model, truncating the flawed reasoning trace at its midpoint, and asking a target model to continue and correct it. We propose the Correct Flip Rate (CFR) as a direct measure of successful recovery and evaluate eight models spanning the Qwen3.5, Gemma4, and Nemotron-3 families. Our results show that even the strongest model achieves a macro-average CFR of only 29.1%, and scale does not reliably improve resilience within families. Nemotron-3-nano 4b outperforms all Qwen3.5 variants including the 27b model and leads the auxiliary TeleMath numerical evaluation at 23.4% CR%, offering the best resilience-to-cost ratio in the set. A difficulty-stratified analysis further reveals that existing telecom benchmark difficulty labels reflect factual specificity rather than reasoning depth, suggesting that current evaluations measure knowledge coverage more than reasoning ability.
title TeleResilienceBench: Quantifying Resilience for LLM Reasoning in Telecommunications
topic Machine Learning
Software Engineering
url https://arxiv.org/abs/2605.09929