Shipping translation systems now means coping with output that may be rewritten, polished, expanded, translated, or lightly edited before it reaches users, so detectors and evaluators have to survive more than the raw model text. The papers here range from matched conditional-response probing for robust LLM-generated text detection to Nemotron 3.5 Super and Lightning, a pair of open MoE hybrid Mamba-Transformer models with 120B/12B-active and 30B/3B-active variants built for reasoning, long context, and agentic workloads.

Robust LLM-Generated Text Detection through Matched Conditional-Response Probing

For edited or translated text, keep a detector that conditions on matched semantic changes, because MCRP stays strong under rewriting, polishing, and expansion and reaches 93.97% F1 there.

Reliable detection of large language model (LLM)-generated text must remain effective under meaning-preserving text revisions, such as rewriting, polishing, and expansion. In practice, LLM outputs are often rewritten, polished, expanded, translated, or lightly edited before publication, weakening the lexical, syntactic, and local discourse cues used by many existing detectors. We introduce Matched Conditional-Response Probing (MCRP), a frozen-LLM probing framework that measures how a candidate text responds when a semantically matched condition is introduced, removed, or replaced. For each candidate, the frozen probe LLM first generates a short semantic condition and then compares token-level losses and prompt--target attention measurements under matched, null, and control conditions. It aggregates bidirectional conditioned alignment, counterfactual sensitivity, and internal attention coupling into a probe representation, on which it trains a lightweight supervised random-forest classifier while keeping the probe LLM frozen. Across standard detection, robustness to meaning-preserving text revisions, multilingual, and multidomain evaluations, MCRP achieves consistently strong performance. Under rewriting, polishing, and expansion, it reaches 93.97% F1 and 96.24% ROC-AUC, outperforming the strongest evaluated baseline by 19.50 and 25.77 percentage points, respectively. Component ablations show that the response feature groups make complementary contributions to predictive performance. In the condition-control study, matched conditions consistently outperform random, mismatched, and corrupted alternatives, demonstrating a measurable incremental benefit from semantic matching.

Nemotron 3.5 Super and Lightning: Open, Efficient Mixture-of-Experts Hybrid Mamba-Transformer Models for Agentic AI

For MT and localisation workloads, Nemotron 3.5 Super’s Indic continued pretraining matters most, since it lifts Indian-language benchmarks by up to 9.82 points and its 3B-active Lightning variant is cheaper to deploy.

We introduce Nemotron 3.5 Super and Nemotron 3.5 Lightning, open and efficient Mixture-of-Experts Hybrid Mamba-Transformer models designed for reasoning, long-context understanding, and agentic workloads. Super is a 120B-parameter, 12B-active multimodal model, while Lightning is a smaller 30B-parameter, 3B-active text model; both designed for efficient deployment. Both models combine large-scale distillation, long-context training, and native speculative decoding, while Super further incorporates Indic continued pretraining and agentic software-engineering mid-training. We develop a comprehensive post-training pipeline spanning supervised fine-tuning and reinforcement learning (RLVR and RLHF), complemented in Super by specialized teacher training and multi- teacher on-policy distillation (MOPD). This transforms Super from a text-only base model into a multimodal agent capable of reasoning over text, images, video, and interactive tool environments. We further explore test-time scaling through parallel agent execution and self-verification. Across training and deployment, we jointly optimize capability and efficiency. Indic continued pretraining improves Indian-language benchmarks by up to 9.82 percentage points, while agentic mid-training strengthens software-engineering capabilities and robustness across agent harnesses. The post-trained Super model achieves 51.78% on SWE-bench Pro, surpassing Nemotron 3 Ultra despite using approximately 4.6x fewer active parameters. Post-training MTP boosting improves speculative- decoding acceptance length by up to 21.8% under probabilistic draft sampling, while NVFP4 quantization achieves 99.62% median accuracy recovery relative to BF16. Inference optimizations enable up to 4,856 output tokens/s/GPU on GB200 while sustaining at least 50 tokens/s/user.

Papers announced by arXiv on Wednesday, 28 October 2026, covering submissions from 27 October to 28 October 2026.