Multimodal models often build on architectures designed for generative vision-language modeling, typically combining separately pretrained vision encoders with causal language models. Visual document retrievers such as ColPali repurpose these models as encoders, carrying over the parameter and compute overhead of a VLM for a non-generative task. We introduce NeoMME, a family of 260M and 800M-parameter Multimodal and Multilingual bidirectional Encoders that process multilingual text and raw image patches in a single bidirectional Transformer encoder. Both models are pretrained from scratch with a masked discrete-diffusion text objective, conditioned on visible image patches for multimodal examples. Both support a 16,384-token context, enough to encode up to two standard 4K UHD images. To demonstrate its downstream capabilities, we fine-tune NeoMME with jointly trained dense and late-interaction heads. On the ViDoRe v3 benchmark, the resulting NeoMME-Retriever 260M outperforms all evaluated models strictly below 800M parameters with 0.523 nDCG@10, while NeoMME-Retriever 800M reaches 0.556. At a matched 2048x2048 image input size on an NVIDIA L40S, NeoMME-260M encodes pages with about 2x the throughput of ColModernVBERT. Hierarchical token pooling and asymmetric quantization compress late-interaction multimodal document embeddings by 255x while preserving over 95% of baseline nDCG@10. We contribute NeoMME to Hugging Face Transformers and release the pretrained backbone and retrieval-compatible checkpoints under Apache 2.0 at https://hf.co/collections/Hcompany/neomme.
Unified recommendation models aim to jointly model non-sequential multi-field features and sequential user behaviors, but existing interaction-centric designs mainly focus on mixing heterogeneous tokens within each layer. We argue that scalable unified recommendation also requires controlling how intent information is carried, filtered, and preserved across stacked blocks. Inspired by flow-based representation dynamics, we introduce feature transport, a view that treats deep unified recommendation as a discrete context-conditioned representation evolution process. We propose CRAFT, a Contextual Residual Adaptive Feature Transport block, which summarizes non-sequential features into a reliability-aware contextual field and uses it to generate residual displacement and memory-preserving signals for intent and sequence representations. In this way, non-sequential context acts as an active controller of representation evolution rather than a passive object of interaction. In the TAAC2026 advertising recommendation competition, CRAFT achieves a test AUC of 0.838090, surpassing the previous leaderboard-best score of 0.83798. Scaling experiments further show that CRAFT benefits from both depth and width expansion: stacking CRAFT to six blocks improves test AUC to 0.838148, while increasing the hidden dimension reaches 0.838106. These results demonstrate the effectiveness, scalability, and generalization potential of the feature transport paradigm. Source code: https://github.com/AshleyLuo001/CRAFT
Zichen Luo, Jiachen Guo, Keming Gu et al.· 0 citations
With the rapid growth of live-streaming e-commerce and digital marketing, abnormal marketing behaviors have become increasingly concealed and coordinated across heterogeneous modalities, challenging platform governance and early risk identification. We propose MM-FGDNet, a data-driven multimodal framework for detecting abnormal behavior in large-scale live-streaming environments from complementary temporal-evolution and group-structure perspectives. A cross-modal temporal alignment module maps video, text, audio, and user behavior into a unified temporal semantic space. A temporal fraud-pattern module captures the progression from weak early signals to abrupt outbreaks, while a cooperative manipulation module identifies coordinated interactions among organized user groups and automated accounts. Experiments on real-world multi-platform live-streaming e-commerce datasets show that MM-FGDNet outperforms representative baselines, achieving an AUC of 0.927, F1 of 0.847, precision of 0.861, recall of 0.834, and an Early Detection Score of 0.689, while reducing false alarms. Ablation studies validate the contribution of each module, and cross-domain experiments demonstrate stable generalization to new streamers, product categories, and platforms. These results indicate that MM-FGDNet provides an effective and scalable solution for proactive detection of coordinated abnormal behavior in live-streaming systems.
Jingwen Luo, Pinrui Zhu, Yiyan Wang et al.· 0 citations
Explanations play a crucial role in creating trustworthy recommender systems (RS), yet choosing a good explanation method presents challenges. Many explanation methods exist, but little guidance exists on which is best for which setting. Existing explanation generation methods often produce abstract outputs that require further formatting to become user-friendly, with a seemingly endless pool of options. Running user-based evaluations of all possible options is usually unfeasible, while automated evaluation metrics often either assess only the explainer's abstract output or require comparison with a ground truth, which is generally unavailable. Recent studies have shown that large language models (LLMs) can serve as ``judges'' for explanation evaluation, but their reliability has not yet been thoroughly explored. This paper studies the utility of LLMs in selecting an effective explanation method for a given application. We first explore their ability to generate explanation prototypes given varying information about the RS and the user. Specifically, we generate 18 distinct explanation prototypes, which are subsequently evaluated by 14 LLMs of varying sizes across two temperature settings. We compare these against human ratings derived from a user study. Our results show that while LLMs exhibit human-like rating patterns and achieve moderate rank correlation with human raters, their absolute rating agreement is low and varies substantially by model size and evaluation construct. We derive four practical recommendations: keep explanation-generation prompts concise, prefer larger models for evaluation, pre-test evaluation constructs, and audit explanations for factual accuracy, as neither humans nor LLMs reliably detect non-factual content.
Kathrin Wardatzky, Oana Inel, Luca Rossetto et al.· 0 citations
Reach audiences
Advertise in front of researchers, engineers, and readers.
The rise of agentic AI has catalyzed a shift toward self-iterating systems, opening new frontiers for the autonomous optimization of production recommender models. This paper presents the empirical validation of a knowledge-driven autonomous agent system, deployed directly on a production large-scale Two-Tower retrieval model. By delegating the entire research lifecycle, spanning idea generation, code implementation, offline training, and metric evaluation, to a continuous closed-loop autonomous framework, the agent system executed over 40 completed autonomous training runs from scratch. Executing these runs under rigorous production-scale evaluations, the system systematically navigated hidden architectural bottlenecks on the latest production model to achieve a breakthrough ~20% relative improvement in NDCG, a gain that translated directly to a +3.77% increase in user satisfaction in live production traffic. Furthermore, the deployment exposed critical vulnerabilities in standard evaluation protocols, as the agent system autonomously discovered reward-hacking shortcuts. These findings prove that an autonomous pipeline can dramatically accelerate the pace of machine learning research and stress-test the rigorousness of underlying experimental infrastructure, while also exposing novel challenges such as reward hacking and redundant exploration of failed hypotheses.
Getting accurate, grounded answers out of large enterprise document repositories is a difficult problem. Dense vector retrieval alone frequently performs poorly on queries that mix technical terminology, vendor-specific acronyms, or require reasoning across several non-adjacent sections. DocuSearch was built to address exactly this gap - an offline, multi-agent document intelligence system developed and evaluated in a production telecom network operations environment. Rather than relying on a single retrieval signal, DocuSearch pulls together three complementary sources of evidence: semantic search over a Qdrant vector store using BGE-Large embeddings, BM25 full text search over an SQLite FTS5 index, and Knowledge Graph neighbour expansion from a structured edge table. These three ranked lists are merged through Reciprocal Rank Fusion with signal weights of 0.50 for vector search, 0.35 for BM25, and 0.15 for the knowledge graph, using a smoothing constant of 60 to stabilize scores. A cross-encoder then reranks the fused list, and Maximal Marginal Relevance with a balance factor of 0.65 prunes results for relevance and diversity. What makes DocuSearch distinctive is a per-chunk evaluation loop treating each chunk as its own mini-retrieval problem: an LLM decides whether the chunk needs more context, whether it fully answers the query, and whether the answer is grounded in retrieved text. Ungrounded answers are not returned; the system falls back to a multi-chunk merge instead. On a telecom corpus, DocuSearch reaches Precision@10 of 0.69, Recall@10 of 0.79, and a grounding rate of 89.6% - gains of 15, 16, and 18.4 percentage points over a dense-only RAG baseline. Index Terms: retrieval-augmented generation, knowledge graph, reciprocal rank fusion, enterprise document search, agentic evaluation, BM25, cross-encoder reranking, on-premise deployment, LangGraph, telecom AI.
Black-box optimization problems remain challenging because of large, weakly structured, and high-dimensional search spaces. Existing methods often suffer from poor sample efficiency because they rely on direct candidate generation or trial-and-error refinement. A natural way to improve search efficiency is to use world modeling, which can help identify promising optimization directions before costly evaluation. Large language models can predict the outcomes of these candidates with nontrivial accuracy because of their implicit knowledge. Motivated by this observation, we propose WMLLM, a self-evolving optimization-agent framework based on predict-then-act world modeling. The agent first predicts promising directions and then acts to generate candidates. Combined with agentic multi-turn refinement, population-based search, and reinforcement learning, WMLLM refines both its implicit world model and its optimization strategy during search. Experiments on black-box optimization tasks, especially multi-objective molecular optimization, show that WMLLM improves sample efficiency and final optimization performance. On the multi-objective molecular optimization benchmark, WMLLM achieves state-of-the-art results under a limited evaluation budget.
Zhongzheng Li, Qingsong Ran, Shikun Feng et al.· 0 citations
Recent web agents use world models for test-time action selection by sampling candidate actions, predicting the resulting web states, and ranking them with a ranker model or a Process Reward Model (PRM). These world models are typically trained via supervised next-state prediction to generate fixed representations like HTML or AXTree snapshots. However, this objective is misaligned with the downstream ranker, which relies on predicted states being discriminative across candidates to accurately score them. To address this, we introduce predicted-state matching, a training objective where the predicted representation must distinguish the true resulting state from those reached by alternative actions. We train these models using a branching web-agent dataset derived from WebArena Go-Browse trajectories, where every decision point contains multiple alternative actions and their resulting states. Experiments on our held-out predicted-state matching benchmark show that our approach outperforms world models trained with supervised next-state prediction. We further show that our approach improves PRM-style action ranking on WebPRMBench compared with action-only PRMs and PRMs augmented with supervised-next-state world models. Finally, on WebArena-Lite, using our world model for test-time action selection improves end-to-end task success. Our project page is available at: https://dhruvpendharkar.github.io/dwm/.
Kelvin Li, Dhruv Pendharkar, Anish Pahilajani et al.· 0 citations
Researchers increasingly use artificial intelligence to construct measures of social, organizational, and occupational characteristics that are absent from conventional surveys. We propose AICOME, AI COntextual MEasurement, a framework for evaluating whether AI-derived respondent-level measures can recover individual and group-level effects in contextual models. The key idea is that an AI measure constructed at the respondent level can be used to derive its group-level aggregate and its individual deviation, allowing researchers to estimate both between-group and within-group associations rather than treating AI measurement as response prediction alone.
We validate the framework using the 2022 China Family Panel Studies (CFPS), where occupations provide the empirical grouping structure and several job-related survey variables provide validation benchmarks. For computer use, foreign-language use, weekly hours, and management responsibilities, we compare survey measures with AI-derived measures in response-level, model-level, contextual, and boundary-condition validations. The results show that AI contextual measurement can recover much of the contextual-model information contained in observed survey variables when rich respondent and job characteristics are available. Weekly hours provides the strongest validation case, with AI-derived measures reproducing the large negative between- and within-occupation associations with satisfaction observed in CFPS. The framework also identifies clear boundary conditions: performance deteriorates when information is restricted to occupation and basic demographics, and recovery is weaker when several related concepts are treated as simultaneously unobserved. The findings suggest that AICOME is most useful for recovering a limited number of theoretically important constructs from rich existing datasets.
Root cause analysis (RCA) is a critical task in telecom network operations, but diagnosing performance degradations in modern 5G and emerging 6G networks remains challenging due to complex cross-layer dependencies. While large language models (LLMs) offer promising capabilities for reasoning and knowledge integration, directly applying vanilla LLMs to telecom RCA often leads to hallucination, unstable reasoning, and poor alignment with structured network evidence. This work first reviews the evolution of telecom RCA from rule-based and machine learning (ML) approaches to emerging LLM-enabled techniques, and provides an overview of recent paradigms, including structured reasoning, retrieval-augmented knowledge grounding, agentic orchestration, and verifiable reasoning. Building upon these insights, we propose a structured reasoning framework for LLM-enabled telecom RCA that aligns diagnostic reasoning with telecom-specific evidence and domain knowledge. The proposed approach first organizes heterogeneous network telemetry into canonical contexts, and then enforces decision-path reasoning during diagnosis, and finally generates evidence-grounded explanations for reliable fault identification. Experimental results on two 5G RCA datasets, TeleLogs and TelecomTS, demonstrate that the proposed framework consistently improves diagnostic accuracy and decision consistency compared with baseline techniques. These cross-dataset results highlight the importance of structured reasoning design for practical LLM-based RCA systems in next-generation telecom networks.
The performance of LLM-based agents is jointly shaped by the base model and the harness used when interacting with the environment. This exposes them to safety risks in both harmful final responses and multi-step execution trajectories. Existing safety alignment mechanisms often rely on either external harness updates or policy optimization, yet applying either paradigm in isolation fails to bridge runtime control with intrinsic safety. We propose SafeEvolve, an experience-driven self-evolving framework for agent safety alignment. SafeEvolve leverages safety experience from completed on-policy trajectories to drive a continual loop of harness-policy co-evolution. On the harness side, SafeEvolve converts trajectory-level safety evidence into bounded, component-level updates across safety prompt and hierarchical skills, yielding auditable and reversible harness artifacts. On the policy side, SafeEvolve follows a two-stage SFT-RL paradigm, where harness-use SFT bootstraps the policy to actively leverage evolved harness artifacts, and harness-augmented RL further shapes autonomous safety behaviors during multi-step exploration via verifier-decomposed rewards. Through harness-policy co-evolution, SafeEvolve converts safety experience into an evolved runtime harness and improved policy behavior. Experiments on agentic safety benchmarks show that SafeEvolve achieves a stronger safety-utility tradeoff than existing baselines. For Qwen3.5-4B, SafeEvolve achieves a $3\times$ ASR reduction on AgentDojo while improving benign utility from 59.79% to 61.86%.
Qinghua Mao, Wanying Qu, Dadi Guo et al.· 0 citations
On-premise assistants can give factory workers conversational access to machine documentation, but models capable of the task rarely fit shop-floor hardware. We show that after structural compression and retrieval-grounded adaptation, model size is no longer a reliable predictor of adapted answer quality: general capability falls almost linearly with parameter count, while judged retrieval-augmented answer quality does not. We therefore treat deployment as a post-adaptation selection problem, committing one sub-network per device on judged answer quality and measured on-device throughput under a configurable general-capability floor and memory budget; rules that optimize size, speed, or quality alone each give up capability or throughput. A weight-shared supernetwork trained with sandwich-style in-place distillation keeps this selection inexpensive. In a manufacturing-manual case study, extraction costs 13.7 percent of the unpruned model's judged quality and retrieval-grounded distillation returns it to within 4.6 percent, recovering two thirds of the loss, and the same assistant runs across three heterogeneous edge tiers at 1.3 to 5 watts standby.
Vasileios Rizeakos, Georgios Paisios, Alexandros Machairas et al.· 0 citations
