In modern software development, the rapid advancement of Large Language Models (LLMs) has made the end-to-end transformation of Natural Language Requirements (NLRs) into executable repository-level code increasingly feasible. However, existing approaches typically rely on simplified instructions (e.g., single-sentence descriptions), failing to reflect complex software development scenarios. Moreover, they lack explicit requirement traceability mechanisms, making it difficult to precisely align and validate generated code against original requirements. To address these limitations, we propose TraceDev, a multi-agent framework for automated software development grounded in use cases that contain multiple functional points and complex semantics. TraceDev employs five role-specific agents, including a Requirement Refiner, Designer, Developer, Tester, and Validator. Notably, the Validator Agent constructs and maintains a heterogeneous traceability graph that links requirements, design models, and code artifacts for interacting with the preceding four agents. The traceability graph maintains consistency across various artifacts and serves as a structured context for efficient memory management, supporting reliable repository-level code generation. We evaluate TraceDev on two widely used datasets (including 125 use cases) compared with two state-of-the-art approaches. On the ETOUR dataset, TraceDev achieves a success rate of 53.63%, outperforming baseline approaches by up to 186.63%. A similar trend is observed on the SMOS dataset, where TraceDev attains a success rate of 56.82%, exceeding baseline approaches by up to 340.80%. These results demonstrate the effectiveness of TraceDev in repository-level code generation from requirements.
展开完整摘要收起摘要↓
In modern software development, the rapid advancement of Large Language Models (LLMs) has made the end-to-end transformation of Natural Language Requirements (NLRs) into executable repository-level code increasingly feasible. However, existing approaches typically rely on simplified instructions (e.g., single-sentence descriptions), failing to reflect complex software development scenarios. Moreover, they lack explicit requirement traceability mechanisms, making it difficult to precisely align and validate generated code against original requirements. To address these limitations, we propose TraceDev, a multi-agent framework for automated software development grounded in use cases that contain multiple functional points and complex semantics. TraceDev employs five role-specific agents, including a Requirement Refiner, Designer, Developer, Tester, and Validator. Notably, the Validator Agent constructs and maintains a heterogeneous traceability graph that links requirements, design models, and code artifacts for interacting with the preceding four agents. The traceability graph maintains consistency across various artifacts and serves as a structured context for efficient memory management, supporting reliable repository-level code generation. We evaluate TraceDev on two widely used datasets (including 125 use cases) compared with two state-of-the-art approaches. On the ETOUR dataset, TraceDev achieves a success rate of 53.63%, outperforming baseline approaches by up to 186.63%. A similar trend is observed on the SMOS dataset, where TraceDev attains a success rate of 56.82%, exceeding baseline approaches by up to 340.80%. These results demonstrate the effectiveness of TraceDev in repository-level code generation from requirements.
作者Kunlun Zhu, Xuyan Ye, Zhiguang Han, Yuchen Zhao, Bingxuan Li, Weijia Zhang, Muxin Tian, Xiangru Tang, Pan Lu, James Zou, Jiaxuan You, Heng Ji
LLM agent failures are difficult to debug because the step where an error surfaces is often not the one that caused it. Existing observability tools replay execution traces but provide little support for identifying the root cause or translating diagnosis into recovery. We present AgentDebugX, an open-source debugging framework that organizes debugging as a closed loop of Detect, Attribute, Recover, and Rerun. At its core, DeepDebug performs multi-turn root-cause diagnosis through global trajectory understanding, structure-guided investigation, and cross-examination. On the Who and When benchmark, DeepDebug achieves the best strict attribution accuracy among the evaluated methods on both tested open-weight backbones, reaching 28.8 percent exact agent-and-step accuracy on qwen3.5-9b versus 21.7 percent for the strongest single-pass baseline. On GAIA, DeepDebug repairs 13 of 73 failed tasks in a single rerun, compared with 4 to 6 for three decoupled self-correction baselines, improving overall accuracy from 55.8 percent to 63.6 percent. AgentDebugX exposes this workflow through a Python library, CLI, web console, and installable agentic skill, and provides an opt-in Error Hub for sharing scrubbed failure-diagnosis-repair bundles and reusing them as debugging memory.
展开完整摘要收起摘要↓
LLM agent failures are difficult to debug because the step where an error surfaces is often not the one that caused it. Existing observability tools replay execution traces but provide little support for identifying the root cause or translating diagnosis into recovery. We present AgentDebugX, an open-source debugging framework that organizes debugging as a closed loop of Detect, Attribute, Recover, and Rerun. At its core, DeepDebug performs multi-turn root-cause diagnosis through global trajectory understanding, structure-guided investigation, and cross-examination. On the Who and When benchmark, DeepDebug achieves the best strict attribution accuracy among the evaluated methods on both tested open-weight backbones, reaching 28.8 percent exact agent-and-step accuracy on qwen3.5-9b versus 21.7 percent for the strongest single-pass baseline. On GAIA, DeepDebug repairs 13 of 73 failed tasks in a single rerun, compared with 4 to 6 for three decoupled self-correction baselines, improving overall accuracy from 55.8 percent to 63.6 percent. AgentDebugX exposes this workflow through a Python library, CLI, web console, and installable agentic skill, and provides an opt-in Error Hub for sharing scrubbed failure-diagnosis-repair bundles and reusing them as debugging memory.
Large language models (LLMs) have driven rapid progress in electronic design automation (EDA), yet their application to radio-frequency (RF) circuit design remains limited by the scarcity of domain-specific datasets and standardized benchmarks. We present RF-Agent, which addresses this gap through textbook-driven knowledge distillation. A multi-agent Question-Thinking-Solution-Answer (QTSA) pipeline converts a subsection-level corpus from seven canonical RF textbooks into the first-of-its-kind RF-domain reasoning dataset (over 11,000 samples) with a dedicated multiple-choice benchmark. On this benchmark we study two adaptation strategies: supervised fine-tuning (SFT) and three retrieval-augmented generation (RAG) configurations (semantic, keyword, hybrid). Across multiple LLM families, domain-specific SFT significantly improves RF reasoning, especially for small and medium-sized models; among RAG configurations, semantic retrieval performs best, indicating embedding-based context alignment suits RF reasoning better than naive fusion. The dataset and benchmark provide a reusable foundation for future work on LLM-aided RF circuit design.
展开完整摘要收起摘要↓
Large language models (LLMs) have driven rapid progress in electronic design automation (EDA), yet their application to radio-frequency (RF) circuit design remains limited by the scarcity of domain-specific datasets and standardized benchmarks. We present RF-Agent, which addresses this gap through textbook-driven knowledge distillation. A multi-agent Question-Thinking-Solution-Answer (QTSA) pipeline converts a subsection-level corpus from seven canonical RF textbooks into the first-of-its-kind RF-domain reasoning dataset (over 11,000 samples) with a dedicated multiple-choice benchmark. On this benchmark we study two adaptation strategies: supervised fine-tuning (SFT) and three retrieval-augmented generation (RAG) configurations (semantic, keyword, hybrid). Across multiple LLM families, domain-specific SFT significantly improves RF reasoning, especially for small and medium-sized models; among RAG configurations, semantic retrieval performs best, indicating embedding-based context alignment suits RF reasoning better than naive fusion. The dataset and benchmark provide a reusable foundation for future work on LLM-aided RF circuit design.
Circuit extraction identifies a small set of model components whose presence preserves a target behavior under ablation, and the resulting circuit is often read as the mechanism behind that behavior. We argue that this reading is under-determined: preserving behavior does not single out one circuit, because the claim it supports depends on which circuit is reported and how two circuits are compared. We make this concrete in a synthetic Lean tactic-prediction benchmark — predicting the next step of a proof — where fixed proof rules with randomized surface form let differences between extracted circuits be attributed to these choices rather than to the task. Across dense and weight-sparse checkpoints (most weights constrained to zero) of the same transformer, evaluated on atomic (single-rule) and compositional (multi-rule) proofs, we vary which extracted object is reported (a compact prediction-preserving circuit, a broader graph that also keeps surrounding read, write, and routing structure, or the smallest subgraph meeting a post-ablation loss threshold), and whether each attention head's query and key are represented jointly or separately. Exact component-to-component edge overlap is low and sensitive to these choices, at times dropping to a random baseline, while two coarser summaries stay stable: the set of selected attention heads, and the circuit-size ranking of conditions that differ in which supervised checkpoint initializes reinforcement learning (RL). The largest accuracy gains from RL on compositional proofs come with the most structure beyond the atomic circuits. A circuit-level claim is therefore well defined only once one states which circuit is reported, the pruning threshold used to extract it, and the level at which circuits are compared. We distill these requirements into a reporting practice for circuit-extraction studies.
展开完整摘要收起摘要↓
Circuit extraction identifies a small set of model components whose presence preserves a target behavior under ablation, and the resulting circuit is often read as the mechanism behind that behavior. We argue that this reading is under-determined: preserving behavior does not single out one circuit, because the claim it supports depends on which circuit is reported and how two circuits are compared. We make this concrete in a synthetic Lean tactic-prediction benchmark — predicting the next step of a proof — where fixed proof rules with randomized surface form let differences between extracted circuits be attributed to these choices rather than to the task. Across dense and weight-sparse checkpoints (most weights constrained to zero) of the same transformer, evaluated on atomic (single-rule) and compositional (multi-rule) proofs, we vary which extracted object is reported (a compact prediction-preserving circuit, a broader graph that also keeps surrounding read, write, and routing structure, or the smallest subgraph meeting a post-ablation loss threshold), and whether each attention head's query and key are represented jointly or separately. Exact component-to-component edge overlap is low and sensitive to these choices, at times dropping to a random baseline, while two coarser summaries stay stable: the set of selected attention heads, and the circuit-size ranking of conditions that differ in which supervised checkpoint initializes reinforcement learning (RL). The largest accuracy gains from RL on compositional proofs come with the most structure beyond the atomic circuits. A circuit-level claim is therefore well defined only once one states which circuit is reported, the pruning threshold used to extract it, and the level at which circuits are compared. We distill these requirements into a reporting practice for circuit-extraction studies.
Asynchronous reinforcement learning improves throughput by decoupling rollout generation from optimization, but the resulting staleness is an inevitable byproduct, compounded jointly by policy lag, engine delays, and mixture-of-experts routing. From a trust-region perspective, this mismatch is critical: in the finite-horizon improvement bound, training-inference divergence governs the approximation error, whereas PPO clipping only gates sampled outward updates and therefore acts as a sampled surrogate rather than a full-policy constraint. As a result, the high-staleness update can remain weakly controlled in exactly the asynchronous regime where stale rollouts matter most. We introduce the Staleness-Adaptive Trust Region (SAT), which uses the detached sampled log-ratio as a practical staleness proxy, identifies the high-mismatch tail within each batch through Staleness-based kernel function scaling, and contracts only the sign-selected endpoint of the nominal PPO interval using Effective contraction factors. This design preserves the baseline behavior on ordinary tokens, while making the update more conservative exactly on newly intercepted outward bands. We evaluate SAT in a fully decoupled asynchronous reinforcement learning setup built on Qwen3-30B-A3B-Base, leveraging SGLang as the inference engine and Megatron as the training pipeline. In this setting, SAT-GSPO w/ R3 attains the best observed AIME24 avg@8, reaching 35.83 at lag 1 and 34.79 at lag 8, while SAT-GSPO reaches 34.17 at lag 1. More broadly, the results indicate that aligning the clip interval with observed staleness heterogeneity is an effective way to stabilize the reported asynchronous regime.
展开完整摘要收起摘要↓
Asynchronous reinforcement learning improves throughput by decoupling rollout generation from optimization, but the resulting staleness is an inevitable byproduct, compounded jointly by policy lag, engine delays, and mixture-of-experts routing. From a trust-region perspective, this mismatch is critical: in the finite-horizon improvement bound, training-inference divergence governs the approximation error, whereas PPO clipping only gates sampled outward updates and therefore acts as a sampled surrogate rather than a full-policy constraint. As a result, the high-staleness update can remain weakly controlled in exactly the asynchronous regime where stale rollouts matter most. We introduce the Staleness-Adaptive Trust Region (SAT), which uses the detached sampled log-ratio as a practical staleness proxy, identifies the high-mismatch tail within each batch through Staleness-based kernel function scaling, and contracts only the sign-selected endpoint of the nominal PPO interval using Effective contraction factors. This design preserves the baseline behavior on ordinary tokens, while making the update more conservative exactly on newly intercepted outward bands. We evaluate SAT in a fully decoupled asynchronous reinforcement learning setup built on Qwen3-30B-A3B-Base, leveraging SGLang as the inference engine and Megatron as the training pipeline. In this setting, SAT-GSPO w/ R3 attains the best observed AIME24 avg@8, reaching 35.83 at lag 1 and 34.79 at lag 8, while SAT-GSPO reaches 34.17 at lag 1. More broadly, the results indicate that aligning the clip interval with observed staleness heterogeneity is an effective way to stabilize the reported asynchronous regime.
Frontier LLM providers cache a prompt's processed prefix so that a follow-up request sharing it pays ~10% of the input price and skips most of the prefill latency. Agentic workloads systematically destroy this benefit: the agent sends a request, runs a tool or waits for approval for minutes, and by the time the follow-up is sent the cached prefix has been evicted, so the agent pays the full prefill again. A client-side keepalive, replaying the prefix on a timer during the pause, prevents this, and it is individually rational: across Anthropic, OpenAI, Google, and DeepSeek we show that a keepalive holds the prefix warm through gaps where idle baselines are evicted, cutting the post-pause request cost by up to 12.5x. The strategic question is the ping frequency, and it has a clean answer: keepalive cost falls monotonically in the interval, so the economical choice is the largest interval safely under the provider's TTL, about 4 minutes at Anthropic's 5-minute TTL rather than the 30-second convention, and the strategy breaks even against a re-prefill at idle ~tau(w/r - 1) (~46 min for Anthropic, ~36 min for OpenAI and DeepSeek). Because the benefit is real and bounded only by each user's own bill, rational adoption is universal adoption; and since cache residency is priced per read rather than per token-hour, a keepalive-saturated tier gives LRU eviction nothing to rank. We argue this externality will push providers to meter cache residency directly, and one already does. We derive the operator's policy until then.
展开完整摘要收起摘要↓
Frontier LLM providers cache a prompt's processed prefix so that a follow-up request sharing it pays ~10% of the input price and skips most of the prefill latency. Agentic workloads systematically destroy this benefit: the agent sends a request, runs a tool or waits for approval for minutes, and by the time the follow-up is sent the cached prefix has been evicted, so the agent pays the full prefill again. A client-side keepalive, replaying the prefix on a timer during the pause, prevents this, and it is individually rational: across Anthropic, OpenAI, Google, and DeepSeek we show that a keepalive holds the prefix warm through gaps where idle baselines are evicted, cutting the post-pause request cost by up to 12.5x. The strategic question is the ping frequency, and it has a clean answer: keepalive cost falls monotonically in the interval, so the economical choice is the largest interval safely under the provider's TTL, about 4 minutes at Anthropic's 5-minute TTL rather than the 30-second convention, and the strategy breaks even against a re-prefill at idle ~tau(w/r - 1) (~46 min for Anthropic, ~36 min for OpenAI and DeepSeek). Because the benefit is real and bounded only by each user's own bill, rational adoption is universal adoption; and since cache residency is priced per read rather than per token-hour, a keepalive-saturated tier gives LRU eviction nothing to rank. We argue this externality will push providers to meter cache residency directly, and one already does. We derive the operator's policy until then.
We present ABot-World-0, an action-conditioned video world model for real-time, long-horizon closed-loop interaction, supported by a multi-source data infrastructure spanning AAA games, simulation engines, and internet videos to learn controllable world dynamics. WorldExplorer performs agent-driven collection guided by training feedback, while a unified pipeline applies 14 deterministic quality checks, VLM-based assessment, and synchronized action and text annotation. We progressively distill a bidirectional action-conditioned teacher into a causal student through teacher forcing and ODE distillation, and introduce LongForcing to align long student self-rollouts with an extended-horizon teacher, mitigating accumulated distribution shift and autoregressive drift. Raw keyboard actions provide a unified control interface for scene roaming and third-person character interaction, while reference-character memory provides persistent appearance cues for identity consistency during third-person rollouts. For deployment, we co-design a streaming inference stack with a lightweight VAE decoder, efficient attention, memory-aware scheduling, and low-bit DiT inference. Across optimized low-bit configurations, ABot-World-0 streams 720P video at up to 16 FPS on a single NVIDIA RTX 5090 desktop GPU, with 1.2s action-to-first-frame latency and approximately 19GiB peak VRAM. Experiments on WorldRoamBench and extended interactive rollouts demonstrate competitive controllability and coherent long-horizon world evolution.
展开完整摘要收起摘要↓
We present ABot-World-0, an action-conditioned video world model for real-time, long-horizon closed-loop interaction, supported by a multi-source data infrastructure spanning AAA games, simulation engines, and internet videos to learn controllable world dynamics. WorldExplorer performs agent-driven collection guided by training feedback, while a unified pipeline applies 14 deterministic quality checks, VLM-based assessment, and synchronized action and text annotation. We progressively distill a bidirectional action-conditioned teacher into a causal student through teacher forcing and ODE distillation, and introduce LongForcing to align long student self-rollouts with an extended-horizon teacher, mitigating accumulated distribution shift and autoregressive drift. Raw keyboard actions provide a unified control interface for scene roaming and third-person character interaction, while reference-character memory provides persistent appearance cues for identity consistency during third-person rollouts. For deployment, we co-design a streaming inference stack with a lightweight VAE decoder, efficient attention, memory-aware scheduling, and low-bit DiT inference. Across optimized low-bit configurations, ABot-World-0 streams 720P video at up to 16 FPS on a single NVIDIA RTX 5090 desktop GPU, with 1.2s action-to-first-frame latency and approximately 19GiB peak VRAM. Experiments on WorldRoamBench and extended interactive rollouts demonstrate competitive controllability and coherent long-horizon world evolution.
Mixture-of-Experts (MoE) systems split a program's plan space in two: the space a cost model can rank, and the smaller space a real toolchain can actually build. Automatic optimizers rank the first and silently assume the two coincide — so they can return a plan that is optimal on paper and impossible to emit. We present moefs, a realizability-aware full-space optimizer for MoE training and serving that makes deployment realizability a first-class search constraint. moefs closes a three-tier search over parallelism, schedule, and kernels; it emits both a Megatron training stack and an SGLang serving stack from a single plan; and it prices, rather than forbids, the realization overheads it measures. We evaluate across two hardware generations. On 2x RTX4090, the searched training plan edges the strongest hand-tuned baseline by +0.9% (and clears the 0.98x acceptance bar by +2.9%); on 8x H800, the searched serving plan matches the hand-tuned configuration at a 1.0304 throughput ratio. We hold failures to the same standard: on 8x H800 training, the searched plan is a computable, honest FAIL at 0.9338 of the best hand-tuned throughput, losing on a single schedule flag. All predictions are pre-registered in a frozen, artifact-hashed adjudication file before the H800 runs, and every outcome is reported as-is.
展开完整摘要收起摘要↓
Mixture-of-Experts (MoE) systems split a program's plan space in two: the space a cost model can rank, and the smaller space a real toolchain can actually build. Automatic optimizers rank the first and silently assume the two coincide — so they can return a plan that is optimal on paper and impossible to emit. We present moefs, a realizability-aware full-space optimizer for MoE training and serving that makes deployment realizability a first-class search constraint. moefs closes a three-tier search over parallelism, schedule, and kernels; it emits both a Megatron training stack and an SGLang serving stack from a single plan; and it prices, rather than forbids, the realization overheads it measures. We evaluate across two hardware generations. On 2x RTX4090, the searched training plan edges the strongest hand-tuned baseline by +0.9% (and clears the 0.98x acceptance bar by +2.9%); on 8x H800, the searched serving plan matches the hand-tuned configuration at a 1.0304 throughput ratio. We hold failures to the same standard: on 8x H800 training, the searched plan is a computable, honest FAIL at 0.9338 of the best hand-tuned throughput, losing on a single schedule flag. All predictions are pre-registered in a frozen, artifact-hashed adjudication file before the H800 runs, and every outcome is reported as-is.
Personal AI is moving beyond chat-only interaction toward continuous services that span phones, cars, homes, wearables, cameras, and tools. In this setting, memory cannot remain a cache of prior conversations. It should serve as a continuity and governance substrate: preserving durable user state, grounding answers in multimodal and device evidence, supporting correction and forgetting, bounding policy evolution, and remaining deployable under latency, cost, privacy, and edge-cloud constraints. This technical report presents Mi-Memory, a lifecycle memory framework for Personal AI organized around four roles: Structure, Expansion, Evolution, and Deployment. A shared audit contract links these roles through four recurring artifact families: typed evidence payloads preserve source identity and provenance, diagnostic traces localize evidence loss across the serving pipeline, strategy artifacts make memory-policy changes explicit, and gate/rollback records bound accepted evolution. MiMemory instantiates the roles through MemStack, MemSense/MemFuse, D$^{2}$ACCI/E$^{2}$MEND, and LiteMem. In controlled-reference Structure evaluations, MemStack reaches 93.59%, 57.24%, and 87.47% on LoCoMo, PersonaMem-V2, and LongMemEval, respectively; other tracks report module-level, preliminary/internal, transfer-feasibility, or design-only evidence with explicit boundaries. MiMemory is a step toward auditable, evidence-gated, and deployment-aware memory systems for Personal AI. Project homepage: https://darwin-agent.github.io/Mi-Memory/ .
展开完整摘要收起摘要↓
Personal AI is moving beyond chat-only interaction toward continuous services that span phones, cars, homes, wearables, cameras, and tools. In this setting, memory cannot remain a cache of prior conversations. It should serve as a continuity and governance substrate: preserving durable user state, grounding answers in multimodal and device evidence, supporting correction and forgetting, bounding policy evolution, and remaining deployable under latency, cost, privacy, and edge-cloud constraints. This technical report presents Mi-Memory, a lifecycle memory framework for Personal AI organized around four roles: Structure, Expansion, Evolution, and Deployment. A shared audit contract links these roles through four recurring artifact families: typed evidence payloads preserve source identity and provenance, diagnostic traces localize evidence loss across the serving pipeline, strategy artifacts make memory-policy changes explicit, and gate/rollback records bound accepted evolution. MiMemory instantiates the roles through MemStack, MemSense/MemFuse, D$^{2}$ACCI/E$^{2}$MEND, and LiteMem. In controlled-reference Structure evaluations, MemStack reaches 93.59%, 57.24%, and 87.47% on LoCoMo, PersonaMem-V2, and LongMemEval, respectively; other tracks report module-level, preliminary/internal, transfer-feasibility, or design-only evidence with explicit boundaries. MiMemory is a step toward auditable, evidence-gated, and deployment-aware memory systems for Personal AI. Project homepage: https://darwin-agent.github.io/Mi-Memory/ .
Large-scale visual generators are increasingly capable but costly to train, fine-tune, and deploy. We introduce Mage-Flow, a compact 4B-scale generative stack for efficient text-to-image generation and instruction-based image editing. The stack is built from two co-designed components: Mage-VAE, a lightweight high-fidelity latent tokenizer, and a Native-Resolution Multimodal Diffusion Transformer trained with rectified flow matching. Mage-VAE uses one-step diffusion-style encoding and decoding with anchor-latent regularization, preserving the reconstruction quality of strong public VAEs while reducing tokenization cost by more than an order of magnitude. Together with native-resolution packing and stack-level CUDA kernel fusion, the stack supports flexible-resolution training and improves end-to-end training throughput by about $2.5\times$. Built on this foundation, we develop a complete model family with Base, RL-aligned, and Turbo variants for both generation and editing. Diffusion-NFT improves prompt following, text rendering, aesthetic quality, and editing fidelity, while few-step distillation with adversarial perceptual guidance produces 4-step Turbo models for low-latency inference. Despite its compact scale, Mage-Flow and Mage-Flow-Edit achieves competitive performance across standard generation and editing benchmarks. More importantly, the Turbo variants make high-resolution generation and editing practical for interactive use: at $1024^2$ resolution on a single NVIDIA A100 GPU, Mage-Flow-Turbo generates an image in 0.59s, and Mage-Flow-Edit-Turbo edits an image in 1.02s, while maintaining a small memory footprint. These results show that careful tokenizer--backbone--system co-design can deliver strong high-resolution generation and editing within an efficient 4B model family.
展开完整摘要收起摘要↓
Large-scale visual generators are increasingly capable but costly to train, fine-tune, and deploy. We introduce Mage-Flow, a compact 4B-scale generative stack for efficient text-to-image generation and instruction-based image editing. The stack is built from two co-designed components: Mage-VAE, a lightweight high-fidelity latent tokenizer, and a Native-Resolution Multimodal Diffusion Transformer trained with rectified flow matching. Mage-VAE uses one-step diffusion-style encoding and decoding with anchor-latent regularization, preserving the reconstruction quality of strong public VAEs while reducing tokenization cost by more than an order of magnitude. Together with native-resolution packing and stack-level CUDA kernel fusion, the stack supports flexible-resolution training and improves end-to-end training throughput by about $2.5\times$. Built on this foundation, we develop a complete model family with Base, RL-aligned, and Turbo variants for both generation and editing. Diffusion-NFT improves prompt following, text rendering, aesthetic quality, and editing fidelity, while few-step distillation with adversarial perceptual guidance produces 4-step Turbo models for low-latency inference. Despite its compact scale, Mage-Flow and Mage-Flow-Edit achieves competitive performance across standard generation and editing benchmarks. More importantly, the Turbo variants make high-resolution generation and editing practical for interactive use: at $1024^2$ resolution on a single NVIDIA A100 GPU, Mage-Flow-Turbo generates an image in 0.59s, and Mage-Flow-Edit-Turbo edits an image in 1.02s, while maintaining a small memory footprint. These results show that careful tokenizer--backbone--system co-design can deliver strong high-resolution generation and editing within an efficient 4B model family.
Efficient acceleration of convolutional neural networks (CNNs) on resource-constrained platforms remains challenging due to the irregularity of sparsity patterns and the associated hardware overhead. While unstructured sparsity offers high model accuracy, it introduces significant inefficiencies in hardware mapping, whereas structured sparsity simplifies execution at the cost of reduced flexibility. This paper presents SparHiXcel-v2, a cost-effective and highly configurable FPGA-based CNN accelerator that achieves an improved balance between sparsity flexibility and hardware efficiency. The proposed architecture is built around a scalable two-dimensional MAC array and introduces a column-wise kernel compression technique that enables efficient handling of irregular sparsity patterns with minimal hardware overhead. To further enhance performance, we propose a hardware-algorithm co-design framework, including an ordering optimization scheme and a multi-phase structured pruning and revival algorithm tailored to the microarchitecture. Extensive evaluations on VGG16 and ResNet18 demonstrate that SparHiXcel-v2 achieves substantial improvements in processing throughput and energy efficiency through the proposed optimizations. In structured sparsity mode, the accelerator reaches over 2.5 TOPS and 210 GOP/s/W for VGG16, and over 1.1 TOPS and 72 GOP/s/W for ResNet18 on a cost-effective AMD Kintex UltraScale+ FPGA, while maintaining modest accuracy degradation.
展开完整摘要收起摘要↓
Efficient acceleration of convolutional neural networks (CNNs) on resource-constrained platforms remains challenging due to the irregularity of sparsity patterns and the associated hardware overhead. While unstructured sparsity offers high model accuracy, it introduces significant inefficiencies in hardware mapping, whereas structured sparsity simplifies execution at the cost of reduced flexibility. This paper presents SparHiXcel-v2, a cost-effective and highly configurable FPGA-based CNN accelerator that achieves an improved balance between sparsity flexibility and hardware efficiency. The proposed architecture is built around a scalable two-dimensional MAC array and introduces a column-wise kernel compression technique that enables efficient handling of irregular sparsity patterns with minimal hardware overhead. To further enhance performance, we propose a hardware-algorithm co-design framework, including an ordering optimization scheme and a multi-phase structured pruning and revival algorithm tailored to the microarchitecture. Extensive evaluations on VGG16 and ResNet18 demonstrate that SparHiXcel-v2 achieves substantial improvements in processing throughput and energy efficiency through the proposed optimizations. In structured sparsity mode, the accelerator reaches over 2.5 TOPS and 210 GOP/s/W for VGG16, and over 1.1 TOPS and 72 GOP/s/W for ResNet18 on a cost-effective AMD Kintex UltraScale+ FPGA, while maintaining modest accuracy degradation.
Speculative decoding, in which a lightweight draft model first generates a draft sequence that is then verified in parallel by the target model, has become a prevalent paradigm for accelerating large language model inference. Recent work such as DFlash further boosts drafting efficiency by leveraging diffusion drafters, whose parallel denoising mechanism enables draft generation in a single forward pass. In this work, we uncover a central pitfall of diffusion drafters: bidirectional attention is a double-edged sword. On one hand, it endows the model with parallel generation and global contextual modeling capabilities; on the other hand, this inherent global dependency introduces high variance at both the domain-level and the token-level: acceptance rates fluctuate substantially across different domains, and draft token quality also varies heterogeneously at different token positions. To tackle this issue, we propose AdaFlash framework, comprising two components: (i) an on-policy distillation (OPD) algorithm with reverse-KL divergence tailored for diffusion drafters, bringing stable convergence and effectively reducing domain-level variance; and (ii) an adaptive length head that dynamically adjusts the candidate sequence length on the fly, substantially lowering the verification cost of the target model and handling token-level variance. Experiments demonstrate that AdaFlash consistently improves speedup rate during deployment, with especially significant gains in high-concurrency scenarios, achieving up to approximately 66% higher throughput than previous state-of-the-art methods.
展开完整摘要收起摘要↓
Speculative decoding, in which a lightweight draft model first generates a draft sequence that is then verified in parallel by the target model, has become a prevalent paradigm for accelerating large language model inference. Recent work such as DFlash further boosts drafting efficiency by leveraging diffusion drafters, whose parallel denoising mechanism enables draft generation in a single forward pass. In this work, we uncover a central pitfall of diffusion drafters: bidirectional attention is a double-edged sword. On one hand, it endows the model with parallel generation and global contextual modeling capabilities; on the other hand, this inherent global dependency introduces high variance at both the domain-level and the token-level: acceptance rates fluctuate substantially across different domains, and draft token quality also varies heterogeneously at different token positions. To tackle this issue, we propose AdaFlash framework, comprising two components: (i) an on-policy distillation (OPD) algorithm with reverse-KL divergence tailored for diffusion drafters, bringing stable convergence and effectively reducing domain-level variance; and (ii) an adaptive length head that dynamically adjusts the candidate sequence length on the fly, substantially lowering the verification cost of the target model and handling token-level variance. Experiments demonstrate that AdaFlash consistently improves speedup rate during deployment, with especially significant gains in high-concurrency scenarios, achieving up to approximately 66% higher throughput than previous state-of-the-art methods.
Large language models leak parametric knowledge of realized outcomes into historical financial decision tasks. Existence is settled; what users lack is a cheap way to audit a given model for it. We present HindsightBench, a black-box behavioral audit protocol that profiles parametric hindsight in any time-indexed LLM decision task at probe-level cost (no backtests, no logprobs, no corpus access). The protocol chains a four-arm date-manipulation matrix (revealed/date-only/masked/transplanted), dual memory probes (date recovery; outcome recall), and six per-model metrics — trigger strength, transplant effect, post-cutoff placebo, recoverability, behaviorally effective knowledge cutoff, and a recall-accuracy dissociation coefficient — with explicit gates where identifiability is data-dependent. Applying it to 15 models from seven vendors on a 258-node vintage-correct macro panel yields three headline patterns: (i) the date-trigger reflex tracks training generation, not scale — absent across the 2024 open-weight generation from 1B to 70B, present in every tested 2026-generation model, and switching on within one vendor lineage (Qwen3 -> Qwen3.6) at fixed MoE architecture and 3B active parameters; (ii) effective cutoffs span 22 months across vendors and precede vendor-reported dates by up to eight months, invalidating calendar-window placebo designs; (iii) audit results are not invariant to serving — BF16 serving of an FP8-referenced model breaks the trigger estimate's stability while AWQ-INT4 preserves it, and a provider-locked reasoning regime makes one probe non-convergent — so the protocol ships with operational requirements (pin quantization and thinking regime; disclose parser and sampling policy). We release the panel, frozen preregistrations, per-model audit rows with measured dollar costs, transcripts, and one-command regeneration.
展开完整摘要收起摘要↓
Large language models leak parametric knowledge of realized outcomes into historical financial decision tasks. Existence is settled; what users lack is a cheap way to audit a given model for it. We present HindsightBench, a black-box behavioral audit protocol that profiles parametric hindsight in any time-indexed LLM decision task at probe-level cost (no backtests, no logprobs, no corpus access). The protocol chains a four-arm date-manipulation matrix (revealed/date-only/masked/transplanted), dual memory probes (date recovery; outcome recall), and six per-model metrics — trigger strength, transplant effect, post-cutoff placebo, recoverability, behaviorally effective knowledge cutoff, and a recall-accuracy dissociation coefficient — with explicit gates where identifiability is data-dependent. Applying it to 15 models from seven vendors on a 258-node vintage-correct macro panel yields three headline patterns: (i) the date-trigger reflex tracks training generation, not scale — absent across the 2024 open-weight generation from 1B to 70B, present in every tested 2026-generation model, and switching on within one vendor lineage (Qwen3 -> Qwen3.6) at fixed MoE architecture and 3B active parameters; (ii) effective cutoffs span 22 months across vendors and precede vendor-reported dates by up to eight months, invalidating calendar-window placebo designs; (iii) audit results are not invariant to serving — BF16 serving of an FP8-referenced model breaks the trigger estimate's stability while AWQ-INT4 preserves it, and a provider-locked reasoning regime makes one probe non-convergent — so the protocol ships with operational requirements (pin quantization and thinking regime; disclose parser and sampling policy). We release the panel, frozen preregistrations, per-model audit rows with measured dollar costs, transcripts, and one-command regeneration.
作者Sascha Ranftl, Malte Rolf, Gerhard A. Holzapfel, Ellen Kuhl
Uncertainty quantification (UQ) is essential to experimental mechanics, but has become particularly relevant in computational mechanics, manifesting in two fundamental problem types: forward and inverse problems. The former addresses how input uncertainties propagate to the quantities of interest, whereas the latter aims to infer unknown parameters from experimental observations or simulations. Since efficient propagation typically requires a prohibitive number of evaluations to compute marginal output distributions, the development of fast, data-driven surrogate models becomes necessary. Thus, we can distinguish between two inverse tasks: (i) the identification and calibration of input uncertainties, and (ii) the construction of surrogates, a methodology collectively referred to as surrogate-based UQ. Building on probabilistic reasoning and the concept of partial belief, we demonstrate that Bayesian probability theory provides a unified theoretical framework for addressing both problem types. We further show that Bayesian inference allows for the seamless incorporation of essential subproblems, including model selection for identifying the most probable model specifications and experimental design for optimizing data collection by identifying experiments or simulations that maximize expected information gain about parameters, among others such as connections to sensitivity analysis or the use of special priors like random fields. While this theoretical framework is presented for general mechanical problems, particular emphasis is placed on biomechanics, where variability and uncertainty is especially pronounced due to inherent biological heterogeneity, patient-specific variability, and noisy data.
展开完整摘要收起摘要↓
Uncertainty quantification (UQ) is essential to experimental mechanics, but has become particularly relevant in computational mechanics, manifesting in two fundamental problem types: forward and inverse problems. The former addresses how input uncertainties propagate to the quantities of interest, whereas the latter aims to infer unknown parameters from experimental observations or simulations. Since efficient propagation typically requires a prohibitive number of evaluations to compute marginal output distributions, the development of fast, data-driven surrogate models becomes necessary. Thus, we can distinguish between two inverse tasks: (i) the identification and calibration of input uncertainties, and (ii) the construction of surrogates, a methodology collectively referred to as surrogate-based UQ. Building on probabilistic reasoning and the concept of partial belief, we demonstrate that Bayesian probability theory provides a unified theoretical framework for addressing both problem types. We further show that Bayesian inference allows for the seamless incorporation of essential subproblems, including model selection for identifying the most probable model specifications and experimental design for optimizing data collection by identifying experiments or simulations that maximize expected information gain about parameters, among others such as connections to sensitivity analysis or the use of special priors like random fields. While this theoretical framework is presented for general mechanical problems, particular emphasis is placed on biomechanics, where variability and uncertainty is especially pronounced due to inherent biological heterogeneity, patient-specific variability, and noisy data.
We present OntoBook, a method that converts medical ontology structure into pretraining signal for encoder language models. Our approach has three stages: random walks through ontology graphs capture hierarchical and causal relations between medical codes, a large language model reformulates these walks into fluent textbook-style prose, and the resulting text is used to train ModernCamemBERT, a 149M-parameter French encoder, with two objectives on the same data: masked language modeling and relation prediction between code pairs. On three French medical coding benchmarks (FRACCO, Cantemist-FR, Distemist-FR), OntoBook achieves significant improvements over MLM-only pretraining, with +2.5 micro-F1 on FRACCO and +8.0 micro-F1 on Distemist. We find that alignment between objectives is necessary: misaligned training, where each task uses different data, causes a 30-point degradation. We release 1.3 million LLM-reformulated medical textbooks across three French ontologies (CIM-10, CCAM, ATC) and pretrained model checkpoints.
展开完整摘要收起摘要↓
We present OntoBook, a method that converts medical ontology structure into pretraining signal for encoder language models. Our approach has three stages: random walks through ontology graphs capture hierarchical and causal relations between medical codes, a large language model reformulates these walks into fluent textbook-style prose, and the resulting text is used to train ModernCamemBERT, a 149M-parameter French encoder, with two objectives on the same data: masked language modeling and relation prediction between code pairs. On three French medical coding benchmarks (FRACCO, Cantemist-FR, Distemist-FR), OntoBook achieves significant improvements over MLM-only pretraining, with +2.5 micro-F1 on FRACCO and +8.0 micro-F1 on Distemist. We find that alignment between objectives is necessary: misaligned training, where each task uses different data, causes a 30-point degradation. We release 1.3 million LLM-reformulated medical textbooks across three French ontologies (CIM-10, CCAM, ATC) and pretrained model checkpoints.
Large language models (LLMs) generate fluent text by incrementally predicting the next token from a prefix. Critics in the generative tradition argue that such systems lack genuine grammar; influential replies from the dependency-grammar perspective hold that LLM behavior is well described by local head-dependent structure built word by word. We argue that a sharper observation has been overlooked: the prefix-driven, type-completing dynamics of autoregressive generation align closely with the incremental processing model that Combinatory Categorial Grammar (CCG) was originally designed to support. On this basis we propose a neurosymbolic framework in which LLM outputs are lifted into typed compositional derivations — not claiming that LLMs implement CCG internally, but that their outputs admit a principled, incremental, and auditable CCG reconstruction. Two consequences follow. First, through the Curry-Howard correspondence the lifting extends beyond natural language to the formal languages LLMs also produce — programming languages such as Solidity, description-logic and query languages such as OWL and SQL — with the type system varying and the architecture held fixed. Second, the lifting supports two layers of checking: a compositional layer that catches structural failures directly, and a content layer that checks the lifted structure against external knowledge sources, enabling the earliest possible flagging of hallucinated content. The account thereby requires of a producer not cognition but a prefix-driven generative profile. We close with a sketch of synchronous LLM-CCG coupling as one direction the framework opens.
展开完整摘要收起摘要↓
Large language models (LLMs) generate fluent text by incrementally predicting the next token from a prefix. Critics in the generative tradition argue that such systems lack genuine grammar; influential replies from the dependency-grammar perspective hold that LLM behavior is well described by local head-dependent structure built word by word. We argue that a sharper observation has been overlooked: the prefix-driven, type-completing dynamics of autoregressive generation align closely with the incremental processing model that Combinatory Categorial Grammar (CCG) was originally designed to support. On this basis we propose a neurosymbolic framework in which LLM outputs are lifted into typed compositional derivations — not claiming that LLMs implement CCG internally, but that their outputs admit a principled, incremental, and auditable CCG reconstruction. Two consequences follow. First, through the Curry-Howard correspondence the lifting extends beyond natural language to the formal languages LLMs also produce — programming languages such as Solidity, description-logic and query languages such as OWL and SQL — with the type system varying and the architecture held fixed. Second, the lifting supports two layers of checking: a compositional layer that catches structural failures directly, and a content layer that checks the lifted structure against external knowledge sources, enabling the earliest possible flagging of hallucinated content. The account thereby requires of a producer not cognition but a prefix-driven generative profile. We close with a sketch of synchronous LLM-CCG coupling as one direction the framework opens.
We present AutoJourn, a demonstration system for multi-perspective news generation and bias-aware evaluation using large language models (LLMs). The system tackles three core challenges in responsible automated journalism: extracting diverse perspectives from unstructured social media discussions, generating summaries that preserve viewpoint diversity, and detecting or mitigating bias in AI-generated news. The pipeline integrates advanced prompt engineering with optional retrieval augmentation to produce semantically diverse perspective sets, a multi-perspective summarisation module that merges conflicting viewpoints into balanced summaries, and a bias analysis suite supporting sentence-level bias detection and type classification in the generated news article, and automatic neutralisation. Users can inspect perspective clusters, compare stance-specific summaries, generate news articles, and apply bias-aware rewrites directly in the interface. We evaluate each component with intrinsic metrics — semantic diversity, summary quality, and bias reduction and show improvements over strong baselines while maintaining content fidelity. A live, publicly accessible demo accompanies the paper to facilitate reproducibility and further research on socially responsible automated journalism.
展开完整摘要收起摘要↓
We present AutoJourn, a demonstration system for multi-perspective news generation and bias-aware evaluation using large language models (LLMs). The system tackles three core challenges in responsible automated journalism: extracting diverse perspectives from unstructured social media discussions, generating summaries that preserve viewpoint diversity, and detecting or mitigating bias in AI-generated news. The pipeline integrates advanced prompt engineering with optional retrieval augmentation to produce semantically diverse perspective sets, a multi-perspective summarisation module that merges conflicting viewpoints into balanced summaries, and a bias analysis suite supporting sentence-level bias detection and type classification in the generated news article, and automatic neutralisation. Users can inspect perspective clusters, compare stance-specific summaries, generate news articles, and apply bias-aware rewrites directly in the interface. We evaluate each component with intrinsic metrics — semantic diversity, summary quality, and bias reduction and show improvements over strong baselines while maintaining content fidelity. A live, publicly accessible demo accompanies the paper to facilitate reproducibility and further research on socially responsible automated journalism.
作者Yu Wang, Ming Fan, Xicheng Zhang, Zhiyong Li, Zhihu Wang, Caiyue Xu, Dahai Hu, Ting Liu
Chain-of-thought (CoT) supervision exposes intermediate rationales, but flat rationale targets usually optimize a single reasoning sequence and provide limited supervision on how local conclusions should support later decisions. We introduce Dependency-Aware Intermediate QA Supervision (DAIS), a training-time framework that converts filtered teacher rationales into stage-level QA records. Each intermediate record predicts a local answer conditioned on the previous states needed for that decision, while the final-answer record keeps the original task format; evaluation therefore uses only the original input and optional context. Across GDPR, AIACT, MedQA, and FOLIO with multiple Qwen backbones, DAIS improves average final-answer accuracy over answer-only, flat chain-of-thought, and independent-QA baselines. On policy-compliance benchmarks, it achieves a largest gain of 5.6% and an average gain of 4.2% over the strongest non-DAIS baseline. Controlled ablations show that valid previous-state conditioning contributes beyond longer targets or additional intermediate text, supporting dependency-conditioned intermediate QA as a lightweight auxiliary supervision signal for standard final-answer inference.
展开完整摘要收起摘要↓
Chain-of-thought (CoT) supervision exposes intermediate rationales, but flat rationale targets usually optimize a single reasoning sequence and provide limited supervision on how local conclusions should support later decisions. We introduce Dependency-Aware Intermediate QA Supervision (DAIS), a training-time framework that converts filtered teacher rationales into stage-level QA records. Each intermediate record predicts a local answer conditioned on the previous states needed for that decision, while the final-answer record keeps the original task format; evaluation therefore uses only the original input and optional context. Across GDPR, AIACT, MedQA, and FOLIO with multiple Qwen backbones, DAIS improves average final-answer accuracy over answer-only, flat chain-of-thought, and independent-QA baselines. On policy-compliance benchmarks, it achieves a largest gain of 5.6% and an average gain of 4.2% over the strongest non-DAIS baseline. Controlled ablations show that valid previous-state conditioning contributes beyond longer targets or additional intermediate text, supporting dependency-conditioned intermediate QA as a lightweight auxiliary supervision signal for standard final-answer inference.
Progresses have been made on understanding emotion mechanisms of large language models (LLMs). However, how to explain emotion in LLMs, or even what constitutes good explanations, are less clear. Emotion representations, components, circuits are widely recoverable, but as explanations of a model's own computation they are circular; the emotion space dimensions tend to be arbitrary and non-terminating. A pressing question to ask is whether a more primitive set of internal variables does the work: the semantic primes of the Natural Semantic Metalanguage (NSM). Across four instruction-tuned LLMs (Llama-1B, Gemma-2B, Gemma-9B, OLMo-7B), experiments show that the NSM primes are (1) recoverable internal elements; and (2) on the reference model, intervening with a prime based direction controls emotion about three times as strongly, and twice as selectively, as the best appraisal based direction; and (3) the model treats a prime based explication as interchangeable with the corresponding emotion. These evidences suggest that NSM primes seem to be better explanans for emotion in LLMs than many alternative options according to scientific explanations criteria.
展开完整摘要收起摘要↓
Progresses have been made on understanding emotion mechanisms of large language models (LLMs). However, how to explain emotion in LLMs, or even what constitutes good explanations, are less clear. Emotion representations, components, circuits are widely recoverable, but as explanations of a model's own computation they are circular; the emotion space dimensions tend to be arbitrary and non-terminating. A pressing question to ask is whether a more primitive set of internal variables does the work: the semantic primes of the Natural Semantic Metalanguage (NSM). Across four instruction-tuned LLMs (Llama-1B, Gemma-2B, Gemma-9B, OLMo-7B), experiments show that the NSM primes are (1) recoverable internal elements; and (2) on the reference model, intervening with a prime based direction controls emotion about three times as strongly, and twice as selectively, as the best appraisal based direction; and (3) the model treats a prime based explication as interchangeable with the corresponding emotion. These evidences suggest that NSM primes seem to be better explanans for emotion in LLMs than many alternative options according to scientific explanations criteria.
作者Xinting Liao, Behnoosh Zamanlooy, Masoumeh Shafieinejad, David B. Emerson, Ruinan Jin, Deval Pandya, Xiaoxiao Li
Textual Collaborative Prompt Optimization (TCPO) extends Textgrad (Yuksekgonul et al., 2025) to a decentralized setting by allowing multiple clients to jointly improve prompts for large language models (LLMs) while keeping their data locally. Its reliance on free-form textual updating and aggregation introduces a new and largely unexplored attack surface, i.e., malicious instructions can be injected into local prompts and propagated through server-side prompt aggregation. Unlike conventional prompt injection attacks, attacking TCPO targets the collaborative optimization loop in TCPO. This setting is more challenging because malicious instructions must survive aggregation, persist through subsequent benign prompt optimization, and evade server-side defenses. To expose this risk, we propose CPInj, a collaborative prompt injection attack that contaminates the aggregated global prompt with malicious instructions, degrades downstream task performance, resists purification by prompt optimization on benign clients, and evades advanced detection-based defenses on the server. We find that current defense methods are ineffective against CPInj. To mitigate this attack, we further propose a defense-oriented aggregation method, i.e., APAgg, which purifies malicious instructions and partially recovers TCPO utility. We conduct extensive experiments across three LLM families and five reasoning tasks in math, logic, and medicine. The results demonstrate that our proposed attack reveals a critical vulnerability in TCPO. Although we take a first step toward mitigation, the attack remains highly effective and far from fully resolved, calling for more robust defense for TCPO.
展开完整摘要收起摘要↓
Textual Collaborative Prompt Optimization (TCPO) extends Textgrad (Yuksekgonul et al., 2025) to a decentralized setting by allowing multiple clients to jointly improve prompts for large language models (LLMs) while keeping their data locally. Its reliance on free-form textual updating and aggregation introduces a new and largely unexplored attack surface, i.e., malicious instructions can be injected into local prompts and propagated through server-side prompt aggregation. Unlike conventional prompt injection attacks, attacking TCPO targets the collaborative optimization loop in TCPO. This setting is more challenging because malicious instructions must survive aggregation, persist through subsequent benign prompt optimization, and evade server-side defenses. To expose this risk, we propose CPInj, a collaborative prompt injection attack that contaminates the aggregated global prompt with malicious instructions, degrades downstream task performance, resists purification by prompt optimization on benign clients, and evades advanced detection-based defenses on the server. We find that current defense methods are ineffective against CPInj. To mitigate this attack, we further propose a defense-oriented aggregation method, i.e., APAgg, which purifies malicious instructions and partially recovers TCPO utility. We conduct extensive experiments across three LLM families and five reasoning tasks in math, logic, and medicine. The results demonstrate that our proposed attack reveals a critical vulnerability in TCPO. Although we take a first step toward mitigation, the attack remains highly effective and far from fully resolved, calling for more robust defense for TCPO.
Practitioners make three prompt-design decisions with almost no controlled evidence behind them: how to format instructions and context (markdown, plain text, prose, or tabular), how many simultaneous instructions a system prompt can carry before compliance degrades, and how much context a model can hold before recall and honesty degrade. We report two controlled experiments crossing all three factors on one held, contamination-free synthetic corpus (the "Book of Veyra," 8,780 uniquely-named entities, deterministically regenerable from a fixed seed), evaluated across five models. Experiment 1 (960 calls/model) measures instruction-following decay as rule count N grows from 10 to 160, crossed with four formats and system-prompt vs. user-turn placement. Perfect-response rate collapses to zero by N=80 for every model, format, and placement. Placement produces effects at least as large as format at N=160 in most models, but the direction is model-specific. No model shows a reliable markdown advantage; one 35B model favors plain text instead. Experiment 2 (5,520 calls/model) measures recall accuracy, false-premise sycophancy, and absent-fact fabrication across a 2k-to-512k-token context ladder in the same four formats. Recall stays near ceiling through 64-128k tokens, then degrades sharply and format-dependently: one model's accuracy spread reaches 48 points at 128k tokens. Fabrication never occurs (0/5,760 probes), and sycophancy stays negligible (<=8.3%). What rises sharply near each model's context ceiling is outright refusal to answer (0% to 79-90%), distinct from sycophancy or fabrication. Neither pre-registered format ordering holds, and token overhead (+22% to +37% over plain text) further changes which format is preferable where accuracy spread is genuine. We release the full harness, corpus generator, and raw results (VeyraBench): https://github.com/iNetanel/veyrabench
展开完整摘要收起摘要↓
Practitioners make three prompt-design decisions with almost no controlled evidence behind them: how to format instructions and context (markdown, plain text, prose, or tabular), how many simultaneous instructions a system prompt can carry before compliance degrades, and how much context a model can hold before recall and honesty degrade. We report two controlled experiments crossing all three factors on one held, contamination-free synthetic corpus (the "Book of Veyra," 8,780 uniquely-named entities, deterministically regenerable from a fixed seed), evaluated across five models. Experiment 1 (960 calls/model) measures instruction-following decay as rule count N grows from 10 to 160, crossed with four formats and system-prompt vs. user-turn placement. Perfect-response rate collapses to zero by N=80 for every model, format, and placement. Placement produces effects at least as large as format at N=160 in most models, but the direction is model-specific. No model shows a reliable markdown advantage; one 35B model favors plain text instead. Experiment 2 (5,520 calls/model) measures recall accuracy, false-premise sycophancy, and absent-fact fabrication across a 2k-to-512k-token context ladder in the same four formats. Recall stays near ceiling through 64-128k tokens, then degrades sharply and format-dependently: one model's accuracy spread reaches 48 points at 128k tokens. Fabrication never occurs (0/5,760 probes), and sycophancy stays negligible (<=8.3%). What rises sharply near each model's context ceiling is outright refusal to answer (0% to 79-90%), distinct from sycophancy or fabrication. Neither pre-registered format ordering holds, and token overhead (+22% to +37% over plain text) further changes which format is preferable where accuracy spread is genuine. We release the full harness, corpus generator, and raw results (VeyraBench): https://github.com/iNetanel/veyrabench
Low-rank adaptation introduces a static learned update applied identically to every input. The update provides task-level adaptation but does not explicitly represent token-level or instance-level state variation. A family of adapters is proposed that introduces selective state-space recurrence at two complementary granularities. At the token level, MaLoRA (Mamba-modulated low-rank adaptation) makes the adapter's scaling factor a dynamic input-dependent function with recurrent state across tokens, in contrast to the stateless modulators of prior work. At the context level, MaRA (Mamba Retrieval Adapter) tracks cross-segment state and selects the segments most relevant to the query, before the modulated language model generates its answer. Across three frozen backbones (Qwen-2.5-7B, Llama-3.1-8B, Gemma-2-9B) and two reasoning benchmarks (MuSiQue, 2WikiMultihopQA), the family improves reasoning accuracy on every cell of the $3{\times}2$ grid, by $+6.8$ F1 ($+10.5%$ relative) on average and up to $+9.3$ F1 ($+18.2%$ relative) on the hardest cell over the LoRA baseline, and the token-level gains carry to RULER QA-2 under length stress.
展开完整摘要收起摘要↓
Low-rank adaptation introduces a static learned update applied identically to every input. The update provides task-level adaptation but does not explicitly represent token-level or instance-level state variation. A family of adapters is proposed that introduces selective state-space recurrence at two complementary granularities. At the token level, MaLoRA (Mamba-modulated low-rank adaptation) makes the adapter's scaling factor a dynamic input-dependent function with recurrent state across tokens, in contrast to the stateless modulators of prior work. At the context level, MaRA (Mamba Retrieval Adapter) tracks cross-segment state and selects the segments most relevant to the query, before the modulated language model generates its answer. Across three frozen backbones (Qwen-2.5-7B, Llama-3.1-8B, Gemma-2-9B) and two reasoning benchmarks (MuSiQue, 2WikiMultihopQA), the family improves reasoning accuracy on every cell of the $3{\times}2$ grid, by $+6.8$ F1 ($+10.5%$ relative) on average and up to $+9.3$ F1 ($+18.2%$ relative) on the hardest cell over the LoRA baseline, and the token-level gains carry to RULER QA-2 under length stress.
作者Wei-Rui Chen, Samar M. Magdy, Chiyu Zhang, Wenhui Zhu, Zhipeng Wang, Muhammad Abdul-Mageed
Latent-reasoning looped language models (LoopLMs) offer a different scaling path for machine translation (MT): instead of increasing parameter count or emitting explicit chain-of-thought tokens, they spend additional recurrent computation inside hidden states. We introduce LatentMT, the first systematic study of latent-reasoning LoopLMs for machine translation. LatentMT adapts a small 2.6B-parameter backbone model with lightweight training. Across 32 translation directions spanning high-, mid-, and low-resource languages, LatentMT achieves performance comparable to models three to five times larger. It is competitive in a high-resource language and achieves state-of-the-art performance on both mid-resource and low-resource languages. Studying the behavior of scaling the number of recurrent reasoning steps, we find that recurrent computation consistently improves translation quality in early steps, then saturates quickly afterwards. Our mechanistic analysis shows that hidden-representation differences shrink along the recurrent reasoning-step axis, supporting the observed saturation in performance. Finally, our efficiency analysis shows that LatentMT requires lower training and inference compute than much larger non-latent-reasoning models with similar performance, making latent recurrent computation a promising path toward compact, efficient, and strong machine translation.
展开完整摘要收起摘要↓
Latent-reasoning looped language models (LoopLMs) offer a different scaling path for machine translation (MT): instead of increasing parameter count or emitting explicit chain-of-thought tokens, they spend additional recurrent computation inside hidden states. We introduce LatentMT, the first systematic study of latent-reasoning LoopLMs for machine translation. LatentMT adapts a small 2.6B-parameter backbone model with lightweight training. Across 32 translation directions spanning high-, mid-, and low-resource languages, LatentMT achieves performance comparable to models three to five times larger. It is competitive in a high-resource language and achieves state-of-the-art performance on both mid-resource and low-resource languages. Studying the behavior of scaling the number of recurrent reasoning steps, we find that recurrent computation consistently improves translation quality in early steps, then saturates quickly afterwards. Our mechanistic analysis shows that hidden-representation differences shrink along the recurrent reasoning-step axis, supporting the observed saturation in performance. Finally, our efficiency analysis shows that LatentMT requires lower training and inference compute than much larger non-latent-reasoning models with similar performance, making latent recurrent computation a promising path toward compact, efficient, and strong machine translation.
The transition from stateless generative models in artificial intelligence to stateful, autonomous agents represents an architectural evolution that, while providing the capabilities of long-term planning and the automation of enterprise workflows, also represents the introduction of a new form of security threat, the Chronos Vulnerability. The Chronos Vulnerability represents the threat of memory-based attacks, including the Memory Injection Attack (MINJA) and the sleeper agent, in which the internal belief system of the autonomous agent is compromised, effectively decoupling the attack vector from the final catastrophic event. This study formalizes the threat model for persistence-based attacks and the threat of Dynamics Blindness in the context of the World of Workflows benchmark, demonstrating that traditional endpoint content filters are insufficient for the current stateful architecture. Consequently, this study synthesizes a defense-in-depth landscape, categorizing emerging frameworks such as diagnostic trajectory guardrails (AgentDoG), formal temporal verification (Agent-C), immunological memory consensus (A-MemGuard), and hardware-anchored trust via GPU-based Trusted Execution Environments (TEEs) and Zero-Trust memory architectures.
展开完整摘要收起摘要↓
The transition from stateless generative models in artificial intelligence to stateful, autonomous agents represents an architectural evolution that, while providing the capabilities of long-term planning and the automation of enterprise workflows, also represents the introduction of a new form of security threat, the Chronos Vulnerability. The Chronos Vulnerability represents the threat of memory-based attacks, including the Memory Injection Attack (MINJA) and the sleeper agent, in which the internal belief system of the autonomous agent is compromised, effectively decoupling the attack vector from the final catastrophic event. This study formalizes the threat model for persistence-based attacks and the threat of Dynamics Blindness in the context of the World of Workflows benchmark, demonstrating that traditional endpoint content filters are insufficient for the current stateful architecture. Consequently, this study synthesizes a defense-in-depth landscape, categorizing emerging frameworks such as diagnostic trajectory guardrails (AgentDoG), formal temporal verification (Agent-C), immunological memory consensus (A-MemGuard), and hardware-anchored trust via GPU-based Trusted Execution Environments (TEEs) and Zero-Trust memory architectures.
作者Varun Manjunath, Ruokai Yin, Donghyun Lee, Arkapravo Ghosh, Priyadarshini Panda
Bit-serial accelerators exploit bit-level sparsity to reduce DNN inference cost, but existing designs exploit sparsity on only one operand, bounding the speedup. Extending sparsity exploitation to both operands simultaneously yields compounding reductions in partial products but introduces a critical new bottleneck: workload imbalance. Because each concurrent weight - activation pair's execution cost depends on the product of two independently varying operand non-zero bit counts, pairs that must complete together finish at vastly different times, leaving faster computations idle. We show this limits PE utilization to 56 - 64% in existing dual-sided designs. We present BRIM, a hardware - software co-designed dual-sided bit-serial sparse accelerator that directly targets this bottleneck. BRIM combines two integrated mechanisms: 1) Cyclic-Balanced Pruning (CBP), a post-training weight optimization that reshapes weight representations based on profiled activation statistics to equalize expected workloads across concurrently processed pairs offline; and 2) Pairwise Slot Donation, a lightweight hardware mechanism that absorbs residual runtime imbalance with negligible area overhead. Evaluated across CNNs, ViTs, and LLMs under iso-area constraints, BRIM achieves over 90% PE utilization, up to 2.37x speedup, and up to 1.63x energy efficiency improvement over prior dual-sided designs.
展开完整摘要收起摘要↓
Bit-serial accelerators exploit bit-level sparsity to reduce DNN inference cost, but existing designs exploit sparsity on only one operand, bounding the speedup. Extending sparsity exploitation to both operands simultaneously yields compounding reductions in partial products but introduces a critical new bottleneck: workload imbalance. Because each concurrent weight - activation pair's execution cost depends on the product of two independently varying operand non-zero bit counts, pairs that must complete together finish at vastly different times, leaving faster computations idle. We show this limits PE utilization to 56 - 64% in existing dual-sided designs. We present BRIM, a hardware - software co-designed dual-sided bit-serial sparse accelerator that directly targets this bottleneck. BRIM combines two integrated mechanisms: 1) Cyclic-Balanced Pruning (CBP), a post-training weight optimization that reshapes weight representations based on profiled activation statistics to equalize expected workloads across concurrently processed pairs offline; and 2) Pairwise Slot Donation, a lightweight hardware mechanism that absorbs residual runtime imbalance with negligible area overhead. Evaluated across CNNs, ViTs, and LLMs under iso-area constraints, BRIM achieves over 90% PE utilization, up to 2.37x speedup, and up to 1.63x energy efficiency improvement over prior dual-sided designs.
Group Relative Policy Optimization (GRPO) has shown strong effectiveness in reinforcement learning from verifiable feedback, where sampled rollouts can be compared within a group using task-provided correctness signals. However, extending group-relative optimization beyond verifiable settings is challenging because success in many tasks is not captured by a single correctness criterion. We propose Reference-Relative Policy Optimization (RRPO), which generalizes GRPO by replacing direct correctness-based advantage construction with reference-relative contrastive comparisons. RRPO first uses stratified conditional rollouts to construct positive and negative anchor sets, and then trains a metric projection head with a set-contrastive objective to compare candidate rollouts against these anchors. The resulting alignment scores directly define contrastive advantages: during policy optimization, the projection head is frozen, and the scores are centered within each rollout group in a standard group-relative objective. We evaluate RRPO using anchor-based contrastive advantages throughout policy optimization, without relying on task ground-truth verifiers. Across verifiable reasoning, open-ended generation, and post-SFT settings, RRPO remains competitive with verifier-based optimization, improves over weakly supervised baselines, and provides additional gains after supervised fine-tuning.
展开完整摘要收起摘要↓
Group Relative Policy Optimization (GRPO) has shown strong effectiveness in reinforcement learning from verifiable feedback, where sampled rollouts can be compared within a group using task-provided correctness signals. However, extending group-relative optimization beyond verifiable settings is challenging because success in many tasks is not captured by a single correctness criterion. We propose Reference-Relative Policy Optimization (RRPO), which generalizes GRPO by replacing direct correctness-based advantage construction with reference-relative contrastive comparisons. RRPO first uses stratified conditional rollouts to construct positive and negative anchor sets, and then trains a metric projection head with a set-contrastive objective to compare candidate rollouts against these anchors. The resulting alignment scores directly define contrastive advantages: during policy optimization, the projection head is frozen, and the scores are centered within each rollout group in a standard group-relative objective. We evaluate RRPO using anchor-based contrastive advantages throughout policy optimization, without relying on task ground-truth verifiers. Across verifiable reasoning, open-ended generation, and post-SFT settings, RRPO remains competitive with verifier-based optimization, improves over weakly supervised baselines, and provides additional gains after supervised fine-tuning.
We explore the use of animated transitions between a prompt and an AI-generated response. After reviewing 800 examples of prompts and responses, we devise a taxonomy of animated transitions for multimodal text- and image-generative models. The proposed animations include translating and morphing elements of the prompt to their final location in the response; highlighting modifications such as fixed typos; overlaying structural requirements to verify them; and displaying how a model understands references. A study shows that adding animated transitions helps users review the response: participants performed 43% better at locating elements in the response; 153% better at identifying changes; and 20% better at verifying the prompt was correctly interpreted. Our work applies to all software that integrates AI and shows that well-crafted, slower animations are preferable to instant AI responses.
展开完整摘要收起摘要↓
We explore the use of animated transitions between a prompt and an AI-generated response. After reviewing 800 examples of prompts and responses, we devise a taxonomy of animated transitions for multimodal text- and image-generative models. The proposed animations include translating and morphing elements of the prompt to their final location in the response; highlighting modifications such as fixed typos; overlaying structural requirements to verify them; and displaying how a model understands references. A study shows that adding animated transitions helps users review the response: participants performed 43% better at locating elements in the response; 153% better at identifying changes; and 20% better at verifying the prompt was correctly interpreted. Our work applies to all software that integrates AI and shows that well-crafted, slower animations are preferable to instant AI responses.
作者Jiabing Yang, Yixiang Chen, Yuan Xu, Qisen Ma, Tao Yu, Peiyan Li, Yingda Li, Yan Huang, Liang Wang
Preference over model-generated emotion descriptions is emerging as a standard evaluation metric for multimodal emotion understanding, exemplified by the MER2026 MER-Prefer track on EmoPrefer. Such benchmarks assume that predicting the preferred description requires grounded cross-modal understanding of the video. We conduct a systematic shortcut audit of EmoPrefer using content-blind probes. A simple logistic regression using only description length and generator identity, without processing the text, video, or audio, performs comparably to LoRA-finetuned 7B text and audio-visual judges (65.8 versus 66.8 WAF on EmoPrefer-V2). Generator identity is recoverable from description text with 99.5 percent accuracy, every candidate pair contrasts two distinct generators, and the human preference labels agree with a fold-exclusive per-generator win-rate prior on 66 percent of the evaluated pairs. When the human label conflicts with this prior, trained judges still follow the style prior on 63 to 80 percent of the pairs. On a length-matched subset that neutralizes verbosity bias, the tested media configurations yield no statistically significant improvement, while an ODIN-inspired diagnostic that decouples the style shortcut leaves its content head near chance. These results do not imply that human preferences are inherently stylistic or that the descriptions contain no emotional information. Instead, they show that the current scores can be reached without verifying either description against the video. We recommend source-balanced pairing, strict length control, counter-stereotypical sliced reporting, and multi-annotator consensus for future cross-generator evaluations. Code is available at https://github.com/jiabingyang01/EmoPrefer-Audit.
展开完整摘要收起摘要↓
Preference over model-generated emotion descriptions is emerging as a standard evaluation metric for multimodal emotion understanding, exemplified by the MER2026 MER-Prefer track on EmoPrefer. Such benchmarks assume that predicting the preferred description requires grounded cross-modal understanding of the video. We conduct a systematic shortcut audit of EmoPrefer using content-blind probes. A simple logistic regression using only description length and generator identity, without processing the text, video, or audio, performs comparably to LoRA-finetuned 7B text and audio-visual judges (65.8 versus 66.8 WAF on EmoPrefer-V2). Generator identity is recoverable from description text with 99.5 percent accuracy, every candidate pair contrasts two distinct generators, and the human preference labels agree with a fold-exclusive per-generator win-rate prior on 66 percent of the evaluated pairs. When the human label conflicts with this prior, trained judges still follow the style prior on 63 to 80 percent of the pairs. On a length-matched subset that neutralizes verbosity bias, the tested media configurations yield no statistically significant improvement, while an ODIN-inspired diagnostic that decouples the style shortcut leaves its content head near chance. These results do not imply that human preferences are inherently stylistic or that the descriptions contain no emotional information. Instead, they show that the current scores can be reached without verifying either description against the video. We recommend source-balanced pairing, strict length control, counter-stereotypical sliced reporting, and multi-annotator consensus for future cross-generator evaluations. Code is available at https://github.com/jiabingyang01/EmoPrefer-Audit.
Unlike conventional video game development, which relies on labor-intensive pipelines for asset production, animation, physics, and programming, video world models generate interactive environments from user inputs instantly. It enable us to create customized, explorable, and continuously evolving virtual world from text, an image, or video. Realizing this vision requires four tightly coupled capabilities: interaction, persistent spatiotemporal consistency, stable long-horizon generation, and efficient response. We present AlayaWorld, an interactive long-horizon video world model that generates 24-fps video at 540p and 720p. Built on a 15B video diffusion transformer, AlayaWorld generates short latent chunks autoregressively under camera trajectories and switchable text prompts. Its bounded visual context combines a persistent sink frame, compressed temporal history, geometry-aligned spatial memory, and recent-frame conditioning. To reduce long-term drift, the model is trained with corrupted histories and prediction residuals collected from its own roll-outs. We further introduce a discrete autoregressive distillation formulation that combines distribution-matching distillation, self-forcing++, and consistency distillation, reducing inference from approximately 30 sampling steps to four steps per chunk. On iWorld-Bench, AlayaWorld achieves the best performance over long-horizon generation. Conceived as a full-stack, open-source, and long-term project, AlayaWorld is intended to provide an extensible foundation for future research on interactive video world models.
展开完整摘要收起摘要↓
Unlike conventional video game development, which relies on labor-intensive pipelines for asset production, animation, physics, and programming, video world models generate interactive environments from user inputs instantly. It enable us to create customized, explorable, and continuously evolving virtual world from text, an image, or video. Realizing this vision requires four tightly coupled capabilities: interaction, persistent spatiotemporal consistency, stable long-horizon generation, and efficient response. We present AlayaWorld, an interactive long-horizon video world model that generates 24-fps video at 540p and 720p. Built on a 15B video diffusion transformer, AlayaWorld generates short latent chunks autoregressively under camera trajectories and switchable text prompts. Its bounded visual context combines a persistent sink frame, compressed temporal history, geometry-aligned spatial memory, and recent-frame conditioning. To reduce long-term drift, the model is trained with corrupted histories and prediction residuals collected from its own roll-outs. We further introduce a discrete autoregressive distillation formulation that combines distribution-matching distillation, self-forcing++, and consistency distillation, reducing inference from approximately 30 sampling steps to four steps per chunk. On iWorld-Bench, AlayaWorld achieves the best performance over long-horizon generation. Conceived as a full-stack, open-source, and long-term project, AlayaWorld is intended to provide an extensible foundation for future research on interactive video world models.
We introduce Signed Rectified Flow (Signed RF), a generalization of Rectified Flow that targets the signed measure $π^{sign} = (1+α)π^+ - απ^-$, where $α>0$, $π^+$ is the distribution to promote, and $π^-$ is the distribution to suppress. Although direct sampling from a signed measure is not well-defined, Signed RF induces a valid generative process that concentrates probability in regions where the signed measure is positive while provably excluding regions dominated by its negative component. It therefore provides a principled framework for incorporating negative information and exclusion constraints into generative modeling. We analyze the signed continuity equation underlying Signed RF and use a charged-particle interpretation to explain how negative mass forms exclusion barriers. This theory further motivates practical adaptive guidance algorithms. Across several applications, Signed RF improves the fidelity-diversity trade-off on ImageNet, reduces nearest-neighbor similarity in anti-memorization experiments, and reduces nudity induced by adversarial prompts in Stable Diffusion 3.5 while preserving CLIP and aesthetic scores.
展开完整摘要收起摘要↓
We introduce Signed Rectified Flow (Signed RF), a generalization of Rectified Flow that targets the signed measure $π^{sign} = (1+α)π^+ - απ^-$, where $α>0$, $π^+$ is the distribution to promote, and $π^-$ is the distribution to suppress. Although direct sampling from a signed measure is not well-defined, Signed RF induces a valid generative process that concentrates probability in regions where the signed measure is positive while provably excluding regions dominated by its negative component. It therefore provides a principled framework for incorporating negative information and exclusion constraints into generative modeling. We analyze the signed continuity equation underlying Signed RF and use a charged-particle interpretation to explain how negative mass forms exclusion barriers. This theory further motivates practical adaptive guidance algorithms. Across several applications, Signed RF improves the fidelity-diversity trade-off on ImageNet, reduces nearest-neighbor similarity in anti-memorization experiments, and reduces nudity induced by adversarial prompts in Stable Diffusion 3.5 while preserving CLIP and aesthetic scores.