Search: “llm”

Prompt Interference in LLM Post-Training: Why Pass@k Optimization Can Degrade Pass@1

arXiv:2602.21189v3 Announce Type: replace Abstract: Pass@k is a widely used performance metric for verifiable large language model tasks, including mathematical reasoning, code generation, and short-a

safetyarxiv-cs-lg

Human

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Model Releases

Refuses in Chat, Complies with Arms: LLM Refusal Does Not Survive Robot Control

DGX agent

arXiv:2610.14043v1 Announce Type: cross Abstract: Frontier language models reliably refuse to explain how to cause physical harm. We ask whether this refusal survives when the same models control a ro

model-releasesarxiv-cs-ro

14 Oct 2026

Requests of a Feather Must Flock Together: Batch Size vs. Prefix Homogeneity in LLM Inference

arXiv:2605.06046v2 Announce Type: replace Abstract: Auto-regressive token generation in large language models is memory-bound because it requires 'attending to' key and value tensors (KV cache) of all

local-aiarxiv-cs-lg

SARATHI: Decoupled Data-Free Probing and Adaptive Reconstruction for Structured LLM Pruning

arXiv:2610.15514v1 Announce Type: new Abstract: Structured pruning reduces the inference cost of large language models (LLMs) by permanently removing neurons and weight rows, yielding dense, hardware-

safetyarxiv-cs-lg

Tracing where LLM 'slop' style comes from in the model and in training, and cutting it out [R]

model-releasesr-machinelearning

I wanted to find where the recognizable assistant style (em dashes, 'it's not X, it's Y', lists of three, buzzwords) lives in a model and which training stage installs it. Where it lives: in Qwen3-4B-

WildTrace: Evaluating Training Data Detection in Realistic Large-Scale LLM Post-Training

arXiv:2610.16152v1 Announce Type: new Abstract: Although post-training significantly enhances model performance, the use of massive undisclosed data raises concerns about benchmark contamination and c

ZOIF: Zeroth-Order Influence Function Estimation for LLM Data Selection Using Only Forward Passes

arXiv:2610.15336v1 Announce Type: new Abstract: Influence functions enable principled data selection for large language models (e.g., identifying influential training samples for model editing, debugg

ABLEIST: Intersectional Disability Bias in LLM-Generated Hiring Scenarios

arXiv:2510.10998v2 Announce Type: replace-cross Abstract: Large language models (LLMs) are increasingly under scrutiny for perpetuating identity-based discrimination in high-stakes domains such as hir

AgentTune: A Unified Training Lifecycle for Tool-Using LLM Agents

arXiv:2610.13335v1 Announce Type: new Abstract: Tool-using agents increasingly require a common substrate for building, training, evaluating, distilling, and maintaining agentic systems, yet existing

COFFEQA: Counterfactual-Guided Few-Shot Distillation for LLM Question Answering

arXiv:2610.12837v1 Announce Type: cross Abstract: Knowledge distillation aims to train a smaller student model to mimic a larger teacher model for resource-constrained environments. However, the proce

researcharxiv-cs-ai

Knowing Is Not Telling: Identity Ossification in LLM Representations of Trans-border Communities

arXiv:2610.12610v1 Announce Type: cross Abstract: Ethnic and cultural identities rarely align with political borders, a tension made sharper by migration and long-standing cross-border ties. Communiti

Matched-Context Group Policy Optimization for LLM Agent Training

arXiv:2610.13755v1 Announce Type: cross Abstract: Training long-horizon language model agents requires assigning credit to intermediate decisions from sparse terminal feedback. Within a finite group o

Mitigating Capability Degradation in LLM Self-Improvement via Representation Geometry Regularization

arXiv:2610.12855v1 Announce Type: new Abstract: On-Policy Self-distillation is a self-improvement paradigm that enables large language models (LLMs) to improve their capabilities by leveraging the sam

Robust Langevin Sampling under State-Dependent Heavy-Tailed Noise with Applications to LLM Fine-Tuning

arXiv:2610.12865v1 Announce Type: new Abstract: Existing convergence guarantees for non-log-concave sampling with stochastic gradient Langevin Monte Carlo (SG-LMC) typically assume globally Lipschitz

researcharxiv-cs-lg

I've been having a lot of fun trying a wide range of LLM post-training experiments using KerasHub + JAX + TPUs. Might open-source some of th…

researchfrancois-chollet--x

On October 11 2026 François Chollet announced that he has been experimenting with large‑language‑model post‑training techniques using KerasHub, JAX, and TPUs. He indicated that portions of his work ma

11 Oct 2026

Agents

Accurate but Not Humble: Evaluating Epistemic Humility in LLM Agents under Knowledge Conflict

arXiv:2610.12360v1 Announce Type: new Abstract: When retrieved evidence contradicts an agent's prior beliefs, does it revise its answer, acknowledge uncertainty, or persist with an incorrect conclusio

agentsarxiv-cs-ai

Characterizing Overconfident Failure in LLM-Based Code Generation

arXiv:2610.11300v1 Announce Type: cross Abstract: Large language models (LLMs) are increasingly used for automated code generation, but generated programs can appear syntactically plausible while stil

local-aiarxiv-cs-ai

Closed-loop evaluation of LLM agents for embedded software development

arXiv:2610.11447v1 Announce Type: cross Abstract: Large language models (LLMs) are increasingly deployed as coding agents that edit files, run builds and tests, inspect execution results, and repair s

Could LLM Watermark Detection be Public?

arXiv:2610.12106v1 Announce Type: cross Abstract: Watermarking large language models is popular for tracing chatbot and agentic outputs, yet detectors remain unreleased since exposing them could let a

Design Creativity Bench: Measuring creativity in LLM-Generated UI

arXiv:2610.11539v1 Announce Type: cross Abstract: As leading LLMs improve on capability evaluations, their limitations in producing creative outputs on design tasks remain insufficiently characterised

From a Prompt to Repertoires: Evolving Functional REpertoires Enable LLM Continual Learning

arXiv:2610.11373v1 Announce Type: cross Abstract: Continual learning remains challenging for large language models, which must enable models to acquire new skills and knowledge without degrading exist

How Narrative Wrapping Affects LLM Refusal: A Cross-Language Benchmark and Defense

arXiv:2610.11005v1 Announce Type: new Abstract: Safety-aligned language models often refuse a harmful request stated directly but answer the same request inside a role-play or narrative wrapper. We me

LLM-powered Query Expansion for Enhancing Boundary Prediction in Language-driven Action Localization

arXiv:2505.24282v2 Announce Type: replace Abstract: Language-driven action localization in videos requires not only semantic alignment between language query and video segment, but also prediction of

local-aiarxiv-cs-cv

Measuring Cultural Alignment Beyond the Average: A Framework for Evaluating Maternal-Health LLM Interactions in Indian Contexts

arXiv:2610.11586v1 Announce Type: new Abstract: Existing evaluation methods for healthcare LLMs primarily assess factual correctness,safety, and fluency, while providing limited insight into whether g

safetyarxiv-cs-cl

Memory Type Varies: Empowering LLM Agents for Long-Term Memory with Diverse Strategies

arXiv:2610.11573v1 Announce Type: new Abstract: The memory capabilities of Large Language Models (LLMs) have garnered increasing attention recently. Despite great success achieved, existing retrieval-

researcharxiv-cs-ai

Use and Disuse: Intent-Structured Experience Consolidation for Memory and Learning in LLM Agents

arXiv:2610.12124v1 Announce Type: new Abstract: The evolution of Large Language Model agents from single-task execution to long-term autonomous operation highlights the critical challenge of transform

Verdict Without the Rule: Diagnosing and Auditing Regulatory Rule Sensitivity in LLM Compliance Systems

arXiv:2610.12313v1 Announce Type: new Abstract: Large language model compliance systems are deployed on the assumption that a verdict depends on the regulatory rule it is given. We test this directly

When Lower Reconstruction Loss Hurts: Distributionally Robust Refinement for Low-Bit LLM Quantization

arXiv:2610.11226v1 Announce Type: new Abstract: Weight-only post-training quantization (PTQ) relies heavily on reconstruction loss minimization to preserve model quality at low precision. We show that

researcharxiv-cs-ai

Why LLM Agents Favor Their Group: Stakes, Observed Norms, and Reputation

arXiv:2610.11008v1 Announce Type: cross Abstract: Language-model agents favor their own group because they have watched their members favor each other. The group label alone does little once the decis

Activation-Aware Weight Tensorization: A Calibration-Time Preconditioner for Tensor-Network LLM Compression

arXiv:2610.10085v1 Announce Type: new Abstract: Post-training tensor-network compression replaces Transformer linear layers with Tensor Train (TT) or Tree Tensor Network (TTN) operators, but standard

CM-DPO: Constraint-Margin Direct Preference Optimization for LLM Planning

arXiv:2610.09219v1 Announce Type: cross Abstract: Direct Preference Optimization (DPO) treats all constraint violations equally: a 1 budget overshoot and a 1,000 overshoot induce the same training sig

safetyarxiv-cs-lg

COPC: Coupled Off-Policy Correction for Asynchronous LLM Reinforcement Learning

arXiv:2610.09597v1 Announce Type: new Abstract: Asynchronous RL accelerates large language model post-training by decoupling rollout generation from optimization, but trains on stale trajectories. Exi

CurveTQ: Rotation-Free Trellis Quantization of LLM Weights via Curvature-Weighted Search

arXiv:2610.09212v1 Announce Type: new Abstract: The best two-bit weight quantizers for large language models, such as QTIP and Proteus, rotate each weight matrix by a random orthogonal transform, whic

researcharxiv-cs-lg

Epistemic Policy Divergence in Multi-Turn LLM Contamination: A Protocol-Gradient Investigation

arXiv:2609.35308v2 Announce Type: replace Abstract: Large language models treat conversation history as unverified context, so false premises injected into prior turns can be adopted as fact, a failur

Applications

Finding the Right Balance: Relevance and Diversity in LLM Retrieval

arXiv:2610.09412v1 Announce Type: cross Abstract: Retrieval diversification is widely available in retrieval-augmented generation (RAG) frameworks, yet prior studies disagree on whether it improves re

applicationsarxiv-cs-cl

GraphOPD: Graph-Augmented On-Policy Distillation for LLM Agents

arXiv:2610.08959v1 Announce Type: new Abstract: On-policy distillation post-trains large language model agents by supplying dense, step-level guidance from a teacher policy when the reinforcement-lear

Applications

LLM Persuasion Is in the Eye of the Evaluation

arXiv:2610.10232v1 Announce Type: new Abstract: Large language models (LLMs) have already been shown to match or exceed human experts in persuasion. While their persuasive capabilities hold promise fo

applicationsarxiv-cs-cl

LoGRA: Scaling LLM Reinforcement Learning with Low-Rank Gradient Sketches

arXiv:2610.06647v2 Announce Type: replace Abstract: Reinforcement learning has greatly advanced the capabilities of large language models, but its memory demands remain a barrier to broader adoption.

Multi-Label Topic Assignment via LLM Distillation: A Comparative Analysis of Generative vs. Discriminative Student Models

arXiv:2610.09063v1 Announce Type: cross Abstract: Multi-label topic assignment for user-generated content (UGC) -- including product reviews and buyer-seller conversations -- poses unique scalability

Rubric Spans are Label Representations: Joint LLM Encoding for Short Answer Scoring

arXiv:2610.09660v1 Announce Type: new Abstract: Automatic Short Answer Scoring (ASAS) requires models that can score student responses against question-specific criteria while remaining efficient and

AlignQuant: Tile-Aligned Mixed-Precision Quantization for Efficient LLM Generation

arXiv:2610.07457v1 Announce Type: cross Abstract: Fine-grained mixed-precision quantization promises efficient large language model inference, but local precision choices can conflict with regular GPU

local-aiarxiv-cs-ai

Beyond Refusal Patterns: Safe-Role Internalization for Robust and Generalizable LLM Safety Alignment

arXiv:2610.07023v1 Announce Type: new Abstract: Large Language Models (LLMs) have achieved remarkable capabilities but remain vulnerable to jailbreak attacks that elicit harmful or unsafe outputs. Exi

Agents

Disentangling Models from Personas in Heterogeneous LLM Simulations

arXiv:2610.07535v1 Announce Type: cross Abstract: Multi-agent simulations with large language models (LLMs) often operate networks of agents with a single base model. This overlooks the inter-model ef

agentsarxiv-cs-ai

Harmful SFT Leaves a Continuous Trace in LLM Checkpoint Updates

arXiv:2610.07518v1 Announce Type: cross Abstract: Safety auditing of post-trained large language models typically relies on model behavior, requiring model execution and depending on the coverage of a

Language Carries the Expert's Impression: Instrument-Anchored LLM Judges Transfer Counseling-Quality Assessment and Beat In-Domain Training

arXiv:2610.08055v1 Announce Type: new Abstract: Automatic assessment of communication quality in dyadic counseling conversations is bottlenecked by data: expert-rated corpora are small and expensive t

researcharxiv-cs-cl

Agents

One Step at a Time: Trading LLM Autonomy for Process Predictability

arXiv:2610.07817v1 Announce Type: cross Abstract: Organizations automating operational processes need more than a correct outcome: they need to predict how a process will run, know which one actually

agentsarxiv-cs-ai

Rethinking Visual Provenance: Detection and Watermarking Across Direct Visual Generation and LLM-Driven Code Rendering

model-releasesarxiv-cs-cv

arXiv:2610.08137v1 Announce Type: cross Abstract: AI systems create images and videos with image/video generation models or by writing code and graphics descriptions that are then rendered. These rout

Tree Navigation Without LLM Summaries: A Matched-Cost Study of Hierarchical Retrieval for Long-Document QA

arXiv:2610.06902v1 Announce Type: new Abstract: Retrieval-augmented generation grounds language models in external context, but for long documents flat top-k retrieval can cluster on a single region a

researcharxiv-cs-cl