LoopUS converts pretrained LLMs into looped latent refinement models via block decomposition, selective gating, random deep supervision, and confidence-based early exiting to improve reasoning performance.
hub Mixed citations
NVIDIA Nemotron 3: Efficient and Open Intelligence
Mixed citation behavior. Most common role is background (44%).
16
Pith papers citing it
Background
44% of classified citations
abstract
We introduce the Nemotron 3 family of models - Nano, Super, and Ultra. These models deliver strong agentic, reasoning, and conversational capabilities. The Nemotron 3 family uses a Mixture-of-Experts hybrid Mamba-Transformer architecture to provide best-in-class throughput and context lengths of up to 1M tokens. Super and Ultra models are trained with NVFP4 and incorporate LatentMoE, a novel approach that improves model quality. The two larger models also include MTP layers for faster text generation. All Nemotron 3 models are post-trained using multi-environment reinforcement learning enabling reasoning, multi-step tool use, and support granular reasoning budget control. Nano, the smallest model, outperforms comparable models in accuracy while remaining extremely cost-efficient for inference. Super is optimized for collaborative agents and high-volume workloads such as IT ticket automation. Ultra, the largest model, provides state-of-the-art accuracy and reasoning performance. Nano is released together with its technical report and this white paper, while Super and Ultra will follow in the coming months. We will openly release the model weights, pre- and post-training software, recipes, and all data for which we hold redistribution rights.
hub tools
citation-role summary
background 4
method 2
baseline 1
dataset 1
other 1
citation-polarity summary
representative citing papers
Four Over Six adaptively scales blocks in NVFP4 quantization to smaller FP4 values, making representable value distributions more uniform and reducing quantization error especially for near-maximal values.
PrivacySIM shows that conditioning LLMs on user personas like demographics and attitudes improves simulation of privacy choices but reaches only 40.4% accuracy against real responses from 1,000 users.
Structured Recurrent Mixers enable algebraic switching between parallel training and recurrent inference representations, yielding higher throughput, concurrency, and training efficiency than comparable linear-complexity models on language tasks.
Priming transfers knowledge from pre-trained Transformers to hybrid SSM-attention models, recovering performance with minimal additional tokens and showing Gated KalmaNet outperforming Mamba-2 on long-context reasoning at 32B scale.
LLMs exhibit Bayesian-like hypothesis updating with strong-sampling bias and an evaluation-generation gap but generalize poorly outside observed data.
Verbalised evaluation awareness in large reasoning models has only small effects on their outputs across safety and alignment tests.
AVISE provides a new framework and automated SET that identifies jailbreak vulnerabilities in language models with 92% accuracy, finding all nine tested models vulnerable to an augmented Red Queen attack.
Unsupervised single-generation confidence calibration for reasoning LLMs via offline self-consistency proxy distillation outperforms baselines on math and QA tasks and improves selective prediction.
Lightning OPD is an offline on-policy distillation method that matches standard OPD performance at 4x efficiency by enforcing teacher consistency between SFT and distillation phases.
Multi-token prediction induces a two-stage reverse reasoning process in Transformers via gradient decoupling, improving planning on synthetic and realistic tasks.
Claw-Eval is a new trajectory-aware benchmark for LLM agents that records execution traces, audit logs, and environment snapshots to evaluate completion, safety, and robustness across 300 tasks, revealing that opaque grading misses 44% of safety issues.
Irminsul recovers up to 83% of prompt tokens above exact-prefix matching and delivers 63% prefill energy savings per cache hit on MLA-MoE models by content-hashing CDC chunks and applying closed-form kr correction.
citing papers explorer
-
LoopUS: Recasting Pretrained LLMs into Looped Latent Refinement Models
LoopUS converts pretrained LLMs into looped latent refinement models via block decomposition, selective gating, random deep supervision, and confidence-based early exiting to improve reasoning performance.
-
Four Over Six: More Accurate NVFP4 Quantization with Adaptive Block Scaling
Four Over Six adaptively scales blocks in NVFP4 quantization to smaller FP4 values, making representable value distributions more uniform and reducing quantization error especially for near-maximal values.
-
PrivacySIM: Evaluating LLM Simulation of User Privacy Behavior
PrivacySIM shows that conditioning LLMs on user personas like demographics and attitudes improves simulation of privacy choices but reaches only 40.4% accuracy against real responses from 1,000 users.
-
Structured Recurrent Mixers for Massively Parallelized Sequence Generation
Structured Recurrent Mixers enable algebraic switching between parallel training and recurrent inference representations, yielding higher throughput, concurrency, and training efficiency than comparable linear-complexity models on language tasks.
-
Priming: Hybrid State Space Models From Pre-trained Transformers
Priming transfers knowledge from pre-trained Transformers to hybrid SSM-attention models, recovering performance with minimal additional tokens and showing Gated KalmaNet outperforming Mamba-2 on long-context reasoning at 32B scale.
-
Hypothesis generation and updating in large language models
LLMs exhibit Bayesian-like hypothesis updating with strong-sampling bias and an evaluation-generation gap but generalize poorly outside observed data.
-
Evaluation Awareness in Language Models Has Limited Effect on Behaviour
Verbalised evaluation awareness in large reasoning models has only small effects on their outputs across safety and alignment tests.
-
AVISE: Framework for Evaluating the Security of AI Systems
AVISE provides a new framework and automated SET that identifies jailbreak vulnerabilities in language models with 92% accuracy, finding all nine tested models vulnerable to an augmented Red Queen attack.
-
Unsupervised Confidence Calibration for Reasoning LLMs from a Single Generation
Unsupervised single-generation confidence calibration for reasoning LLMs via offline self-consistency proxy distillation outperforms baselines on math and QA tasks and improves selective prediction.
-
Lightning OPD: Efficient Post-Training for Large Reasoning Models with Offline On-Policy Distillation
Lightning OPD is an offline on-policy distillation method that matches standard OPD performance at 4x efficiency by enforcing teacher consistency between SFT and distillation phases.
-
How Transformers Learn to Plan via Multi-Token Prediction
Multi-token prediction induces a two-stage reverse reasoning process in Transformers via gradient decoupling, improving planning on synthetic and realistic tasks.
-
Claw-Eval: Towards Trustworthy Evaluation of Autonomous Agents
Claw-Eval is a new trajectory-aware benchmark for LLM agents that records execution traces, audit logs, and environment snapshots to evaluate completion, safety, and robustness across 300 tasks, revealing that opaque grading misses 44% of safety issues.
-
Irminsul: MLA-Native Position-Independent Caching for Agentic LLM Serving
Irminsul recovers up to 83% of prompt tokens above exact-prefix matching and delivers 63% prefill energy savings per cache hit on MLA-MoE models by content-hashing CDC chunks and applying closed-form kr correction.
- Shepherd: A Runtime Substrate Empowering Meta-Agents with a Formalized Execution Trace
- When LLMs Stop Following Steps: A Diagnostic Study of Procedural Execution in Language Models
- SPEED-Bench: A Unified and Diverse Benchmark for Speculative Decoding