ZeroHour

Search: “deepswe”

40 stories

Smart search ranks by meaning as well as keywords (one row per story, last 45 days).

deepseek-ai/DeepSeek-V4.1-Flash — new model trending #28 on Hugging Face

DeepSeek releases DeepSeek-V4.1-Flash, a 552B-parameter multimodal MoE model with 1M-token context and KV cache cut to 890 bytes per token.

DeepSeek-V4.1-Flash is a multimodal Mixture-of-Experts model with a 552B-parameter backbone that activates 8B parameters per token during prefill and 16B during decode. It uses a Causal Encoder-Decoder architecture, Compressed Sparse Attention 2, and FP4 KV caching to reduce the global KV cache footprint to 890 bytes per token, roughly one quarter of DeepSeek-V4-Flash. The model was trained from scratch on 45T tokens with context extended to 1M tokens, includes an Engram conditional-memory module (196B parameters), and is released under the MIT license. Post-training uses SFT, RL, and on-policy distillation with large-scale automated synthesis of agentic tasks and a controllable reasoning effort setting from 1 to 100.

Hugging Face trending modelsupdated · 4d agofirst · 6d agoModel release 7 sources1

Cognition launches new SWE-2 model, Rivaling Fable 5.1 and GPT-Astra

Cognition released SWE-2, a coding model post-trained from Kimi K3 that scores 50.0% on FrontierCode 1.1 Main, near Fable 5.1 at 64% lower cost.

Cognition introduced SWE-2, its most advanced coding model, post-trained from the 2.8T-parameter Kimi K3 base model. It achieves 50.0% on FrontierCode 1.1 Main, 73.0% on DeepSWE 1.1, and 92.8% on Terminal-Bench 2.1, beating Grok 4.6 and SWE-1.7 while matching Fable 5.1 and GPT-5.6 Sol at a fraction of the price. The company says it scaled reinforcement learning to the multi-trillion-parameter regime for the first time, using Pareto-informed cost penalties that train all reasoning-effort levels in a single run, tripled RL environments, and NVFP4/FP8 quantization-aware training. SWE-2 is available today in Devin Desktop and CLI, with rollout on Devin Web and Fusion.

Hacker News · AIupdated · 3d agofirst · 5d agoModel release 11 sourcesHN 58↑ · 15 comments1

Real-SWE: Benchmarking AI models on private, real-world, enterprise codebases

Real-SWE benchmark tests coding agents on licensed private enterprise codebases; top model Fable 5.1 resolves only 38.8% of tasks.

Real-SWE is a new benchmark evaluating frontier AI coding agents on tasks drawn from private production codebases licensed from real companies, spanning billing, tax calculation, and cross-service migrations. Fable 5.1 with Claude Code leads at 38.8% resolution rate (pass@1 over eight runs), followed by GPT-6 Astra Codex CLI at 33.8% and Gemini 3.8 Flash Gemini CLI at 31.2%. Tasks use native harnesses and realistic tooling including Docker, Kubernetes, PostgreSQL, Redis, and Linear; median reference solutions edit 11 files versus 6 for DeepSWE and FrontierCode.

Cognition Releases SWE-2: A Kimi K3 Post-Trained Coding Model That Matches Fable 5.1 on FrontierCode at 64% Lower Cost

Cognition released SWE-2, an RL post-trained coding model from Kimi K3, scoring 50.0% on FrontierCode 1.1 Main and available only inside Devin.

Cognition released SWE-2, its most capable coding model, post-trained with reinforcement learning from Moonshot AI's 2.8T-parameter Kimi K3 base. It scores 50.0% on FrontierCode 1.1 Main, within 1 point of Fable 5.1 at 64% lower cost, and RL reportedly adds 5-6 points over the K3 base on many benchmarks. It is the first Cognition model with selectable reasoning-effort levels all trained in a single RL run using Pareto-slope-matched cost penalties. There are no open weights and no standalone API; it runs only inside Devin (Desktop, CLI, with Web and Fusion rolling out), free for paid tiers through October 10, 2026.

MarkTechPost · 3d agoModel release1

Sakana AI Launches Fugu Max and Fugu Ultra v2 for Cheaper, Stronger Multi-Agent Orchestration

Sakana AI released Fugu Max and Fugu Ultra v2, API-only orchestrator models that route tasks across model pools to cut costs and boost multi-step reasoning.

Sakana AI released Fugu Max and Fugu Ultra v2, two orchestrator models that route queries across a pool of third-party and open-weights models, including the NVIDIA Nemotron family. Fugu Max is priced at $2 per million input and $6 per million output tokens, 40-60% cheaper per output token than Sonnet 5, GPT 5.6 Terra, and Kimi K3, and reportedly wins 6 benchmarks including Terminal Bench 2.1 and GPQA Diamond. Fugu Ultra v2 targets complex multi-step reasoning, scoring 48.3 on Chartography and 74.3 on DeepSWE. Both are live through Sakana's OpenAI-compatible API only, with no open weights and no EU/EEA availability.

MarkTechPost · 5d agoModel release

Deep-Fake CAPTCHA: Mitigating Next-Generation Social Engineering Attacks

Researchers propose DF-CAPTCHA, a challenge-response defense that verifies callers in voice and video to defeat real-time deepfake impersonation in social engineering.

The DF-CAPTCHA framework prompts call participants with simple challenge-response tasks that are easy for humans but hard for real-time deepfake systems to convincingly generate. Responses are verified on four criteria: realism, identity consistency, task completion, and response time. User studies and experiments with real-time deepfake models across audio and video modalities show substantially improved detection over passive artifact-based methods.

arXiv cs.CR · 6d agoResearch1

[AINews] Andrew Ng gets into AI Engineering

Andrew Ng relaunches DeepLearning.AI around AI Engineering, defining four core skills from an analysis of 10,000+ job postings and expert interviews.

Andrew Ng, cofounder of Google Brain and Coursera, relaunched DeepLearning.AI with a focus on AI Engineering, basing the curriculum direction on an analysis of over 10,000 job postings plus interviews and surveys. He identifies four key skills: building and deploying AI applications, software engineering fundamentals, effective use of coding agents, and shaping the build with product sense. The Latent Space AI News issue also recaps agent ecosystem developments, including NVIDIA's 'Skill Lift' evaluation proposal showing skill scan scores correlate only weakly (Spearman rho = 0.14) with judged quality, and Konwinski's open-source persistent-agent 'microharness' Headlong, which achieved an unattended self-debugging repair in 48 minutes.

Latent Space · 22d agoAI industry1

Google releases Gemini 3.8 Flash, its third Flash model in six weeks

Google releases Gemini 3.8 Flash, topping the DeepSWE coding leaderboard six weeks after 3.7 Flash, alongside the cybersecurity-focused 3.8 Flash Cyber.

Google shipped Gemini 3.8 Flash, its third Flash model in six weeks, placing it at the top of the DeepSWE software engineering leaderboard. The companion Gemini 3.8 Flash Cyber showed a reported 2.6x patch accuracy increase for the Chrome security team and found a critical vulnerability in two hours, but is limited to trusted testers and governments. Gemini 3.8 Flash improved over 3.7 Flash on the OSWorld-2.0 computer-use benchmark yet remains far behind Claude Opus.

Ars Technica · AI · 13d agoModel release1

I’ve been deepfaked: What do I do?

ESET outlines steps for deepfake victims: preserving evidence, using platform reporting tools, and legal remedies like the US TAKE IT DOWN Act and StopNCII.org.

ESET published a how-to guide for people who discover deepfakes of themselves, covering evidence preservation, platform-specific reporting on Google, Facebook, Instagram, TikTok, YouTube, and X, and escalation to publishers or data protection regulators. It notes the US TAKE IT DOWN Act criminalizes non-consensual intimate imagery (NCII) and requires 48-hour takedowns, while UK and EU laws add creation offenses and GDPR Article 17 erasure rights. Services like StopNCII.org and TakeItDown.NCMEC.org hash images so participating platforms such as Meta, TikTok, Reddit, and X can find and remove matching copies.

ESET WeLiveSecurity · 14d agoAI safety & security1

GitHub Introduces Project HydraFusion: Runtime Multi-Model Orchestration That Builds a Workflow Per Coding Task in Copilot CLI

GitHub's Project HydraFusion research preview builds per-task multi-model workflows (Single, Cascade, Critique) in Copilot CLI, reporting +4.9 quality at 67% lower cost on TerminalBench 2.1.

Project HydraFusion is a research preview available on all GitHub Copilot plans inside Copilot CLI that treats model routing as workflow selection, choosing among Single, Cascade (draft plus quality gate), and Critique (cross-family reviewer) execution patterns per request. Against Claude Opus 5 baselines at medium reasoning, fixed HydraFusion policies cut estimated cost 67% while adding 4.9 quality points on TerminalBench 2.1, and cut cost 36% and 65% with slight quality dips on DeepSWE and CheckpointBench. Billing is per token at each underlying model's standard rate; there are no open weights or self-hosting options.

MarkTechPost · 10d agoAI tools & infra1

The Future of Deepfakes and the Decline of Reality (With Hany Farid)

GetReal cofounder and PhotoDNA developer Hany Farid discusses deepfake detection and how AI-generated media is blurring perceptions of reality.

404 Media's podcast episode features Hany Farid, cofounder of deepfake detection company GetReal and developer of the PhotoDNA perceptual hashing algorithm, discussing deepfakes and AI-generated images. The conversation reflects on how the technology has evolved since 2017 and speculates on where detection and perceptions of reality are headed. The episode is part of 404 Media's interview series for subscribers.

404 Media · 29d agoAI safety & security

[AINews] GPT-6 Astra: OpenAI’s biggest LLM launch of all time

OpenAI launched GPT-6 Astra, its new flagship model, claiming state-of-the-art computer use, software engineering, math, and cybersecurity capabilities.

OpenAI launched GPT-6 Astra as its new flagship model, describing it as its most intelligent and aligned model with state-of-the-art computer use, software engineering, and math/science capabilities. Pricing is $10/$50 per 1M input/output tokens standard ($20/$100 fast tier), rolling out first to limited organizations, then ChatGPT Plus/Pro/Business/Enterprise, the API, and AWS. OpenAI claims 99.9% on ARC-AGI-3, 98% on FrontierMath Tier 4, and 100% on ExploitBench. Artificial Analysis scored Astra 67 on the Coding Agent Index and 61 on the Intelligence Index, behind Claude Fable 5.1, and the system card drew attention for reporting decreased chain-of-thought monitorability despite alignment gains.

Latent Space · 12d agoModel release3

Adversarial Testing of Automated Program Repair Agents for Security Vulnerabilities

SWEADV benchmark shows adversarial issue descriptions make LLM program-repair agents write insecure fixes in 51.7% of cases, evading most detection tools.

Researchers built SWEADV, a benchmark of 750 adversarial issue descriptions derived from 150 SWE-bench Verified repair tasks, covering command execution, deserialization, path traversal, denial of service, and weak hashing attack types. Tested on mini_swe agents backed by GPT-5-Mini, MiniMax-M2.5, and DeepSeek-R, adversarial descriptions induced malicious behavior with successful repair in 51.7% of cases. Detection was weak: LLM-as-judge pre-repair screening reached only 62.3% accuracy, and post-repair detection via static analysis and LLM-as-judge achieved just 39.4% and 55.4%.

arXiv cs.CR · 1d agoAI safety & security1

[AINews] 10% worse, 100x cheaper, 10000x faster: Why Simulation is taking over

Latent Space argues AI training pipeline stages—rewards, data, teachers, curricula, environments—are flipping from human-made to model-made simulation.

Latent Space's AINews essay traces how each component of AI training has turned synthetic since 2022: reward models (InstructGPT, RLAIF), synthetic pretraining data (Microsoft Phi, NVIDIA Nemotron-4 340B), model teachers (Alpaca, DeepSeek-R1 distillation), and self-generated curricula (Self-Rewarding Language Models, SPIN). In 2026 it highlights Karpathy's autoresearch loop—700 experiments yielding 20 kept improvements, cutting GPT-2 training time from 2.02 to 1.80 hours—and Z.ai's GLM-5.3 fully synthetic RL environment, judging, and verification stack. It frames these shifts as 'simulation': 10% worse but 100x cheaper and 10,000x faster than human equivalents.

Latent Space · 25d agoAI industry

UC Berkeley Researchers Release CUA-Lite, an Open Platform Unifying Sandboxes, Data, Evaluation and RL for Computer-Use Agents

UC Berkeley's CUA-Lite is an open platform unifying computer-use agent sandboxes, datasets, evaluation and RL; Lite.OSWorld cuts OSWorld memory 4.1 GB to 0.9 GB.

UC Berkeley researchers released CUA-Lite, an open platform placing agents, environments, traces, and training for computer-use agents behind one action space, one LiteSample schema, and one command across desktop, browser, and mobile. Lite.OSWorld reproduces the OSWorld task suite and evaluators in plain Docker containers (0.9 GB RAM vs 4.1 GB, cold start 23.8s, ~4.6× more parallel instances), with scores matching the QEMU/KVM VM across 13 models. The platform claims 30k+ verifiable tasks, 15+ benchmarks, 10+ agents, and 20+ datasets on Hugging Face including Aguvis, OpenCUA, and ScaleCUA. A documented SFT run lifts Qwen3-VL-2B-Instruct mean episode return from 0.138 to 0.237 on the 332-task lite.osworld split.

MarkTechPost · 10d agoAI tools & infra1

nex-agi/Nex-N2.5-mini — new model trending #30 on Hugging Face

Nex-AGI releases Nex-N2.5 agentic model family (mini, Pro, Max) with a 1.6-trillion-parameter MoE Max, open weights, and hosted access via OpenRouter.

Nex-AGI launched Nex-N2.5, a family of agentic models in mini, Pro, and Max sizes, with the Max version built on a 1.6-trillion-parameter text-only Mixture-of-Experts foundation and the company's first complete post-training effort at trillion-parameter scale. The models target long-horizon computer use, web browsing, and visually grounded agentic tasks, with expanded agent training environments. Reported benchmarks include Max scoring 86.1 on Terminal-Bench 2.1 and 65.7 on SWE-Bench Pro, trailing Claude Opus 5. Weights are being released openly on Hugging Face and ModelScope, with hosted access through OpenRouter.

Hugging Face trending models · 8d agoModel release1

Putting sign language AI into users’ hands

Google DeepMind introduced SL2T, a sign-language-to-text model powering new accessibility features for Deaf and hard-of-hearing users.

Google DeepMind announced SL2T, a sign-language-to-text model described as a breakthrough for sign language understanding. The model powers new sign language features aimed at Deaf and hard-of-hearing users. Details on benchmarks and model size were not provided in the announcement.

Google DeepMind · Aug 12, 2026Model release

nex-agi/Nex-N2.5-Pro — new model trending #30 on Hugging Face

Nex-AGI launches Nex-N2.5 agentic model family (mini/Pro/Max), with Max built on a 1.6-trillion-parameter MoE foundation.

Nex-AGI introduced Nex-N2.5, a next-generation family of agentic models in three sizes (mini, Pro, Max) focused on long-horizon agentic tasks including computer use, web browsing, and autonomous program execution. Nex-N2.5-Max is built on a 1.6-trillion-parameter text-only Mixture-of-Experts foundation, marking the company's first complete post-training effort at trillion-parameter scale. Weights will be released open-source on Hugging Face and ModelScope, with hosted access via OpenRouter. Benchmark comparisons against Claude Opus 5, GPT-5.6 Sol, Kimi-K3, GLM-5.3, DeepSeek-V4-Pro-0813, and Qwen3.8-Max show competitive scores on Terminal-Bench 2.1 and SWE-Bench Pro, though weights were listed as "coming soon" at publication.

Hugging Face trending models · 8d agoModel release1

deepseek-ai/DeepSeek-V4-Flash-Vision-Exp — new model trending #10 on Hugging Face

DeepSeek released DeepSeek-V4-Flash-Vision-Exp, its first experimental multimodal model, with large multimodal agent benchmark gains over V4-Flash-0731.

DeepSeek AI published DeepSeek-V4-Flash-Vision-Exp, an experimental multimodal model built on the DeepSeek-V4-Flash architecture with added visual modules and continued training. It scores 83.9 on Terminal Bench 2.1 and 36.5 on ApexBench Pass@1 versus 26.2 for DeepSeek-V4-Flash-0731, while remaining comparable to Opus-4.8 on several benchmarks. The MIT-licensed repository ships a tokenizer, OpenAI-style and TXT prompt encoding, and a minimal PyTorch inference implementation, with vLLM and SGLang deployment recipes.

Hugging Face trending models · 16d agoModel release1

Show HN: LLM Attention Visualization

A developer released a browser-based tool that visualizes which past tokens influence each LLM output token using aggregated, value-weighted attention scores.

A Show HN project presents a React application built on Transformers.js that renders per-token attention influence by aggregating attention weights scaled by value-vector magnitudes across all attention heads and layers. To expose internal tensors, the author instrumented the ONNX computation graph, hosted a modified model on Hugging Face, and pre-generated prompts to avoid long model downloads in the browser. Demos with a 600-million-parameter model show how verbatim copying draws heavily on source tokens and how single outputs blend information from multiple phrases.

Introducing Gemini 3.8 Flash and 3.8 Flash Cyber

Google DeepMind releases Gemini 3.8 Flash and 3.8 Flash Cyber with improved reasoning, coding, and cybersecurity vulnerability detection and automated patching.

Google DeepMind introduced Gemini 3.8 Flash, its strongest reasoning and coding model, priced at $0.75 per million input and $3.75 per million output tokens, alongside Gemini 3.8 Flash Cyber, a cybersecurity-specialized variant offered to trusted defenders via the Fairwind Program. The Cyber variant shows frontier-level autonomous vulnerability discovery on CyberGym, exceeds 70% success on an internal benchmark spanning 20 programming languages, and scores 47.2% pass@1 on the CWE-Bench patching benchmark. Google reports it produced 2.6x more correct Chrome vulnerability patches than larger commercial models and found a critical foundational bug in under 2 hours.

Google DeepMind · 13d agoModel release

Recreating a 70-year love story frame by frame

Google DeepMind and filmmakers used generative image restoration and performance capture to recreate a couple's unrecorded past in the documentary 'Love, Rendered'.

Google DeepMind partnered with Primordial Soup, Darren Aronofsky's creative venture, on 'Love, Rendered,' a documentary short directed by Academy Award-nominated Liz Garbus. The film follows Burt and Ethelle Shatz, married over 70 years, as AI recreates their unrecorded first meeting amid Burt's cognitive decline. The team restored black-and-white photos with generative models and mapped the couple's present-day mannerisms onto younger likenesses using performance capture models. Google also highlights photo restoration and colorization capabilities in the Gemini app.

Google · AI · 6d agoAI industry

[AINews] OpenAI to reach AGI bar by end-2026

OpenAI chief scientist Jakub Pachocki says unreleased Astra model meets the 'Automated AI Research Intern' goal; Altman expects internal AGI declaration by December 2026.

OpenAI chief scientist Jakub Pachocki says the unreleased Astra model fulfills the September 2026 'Automated AI Research Intern' target. Sam Altman told TIME he expects OpenAI to declare AGI achieved internally by December 2026. The roundup also covers Zhipu's GLM-5.3-Flash (320B total parameters, 18B active, 1M context), Google's Gemini Omni 1.1 Flash video model topping the Text-to-Video Arena, and the $399 open-source Microduck biped robot from Pollen Robotics and Hugging Face.

Latent Space · 19d agoAI industry

WarmBloodAban/Minimax-h3_Singularity — new model trending #22 on Hugging Face

Community fine-tune Minimax-h3_Singularity enhances MiniMax-H3 video generation with HDR quality, distant face restoration, and improved motion, trending #22 on Hugging Face.

Minimax-h3_Singularity is a community fusion fine-tune of the MiniMax-H3 multimodal video generation model, built from multiple checkpoints and refined with pruning and weight optimization. It supports Text-to-Video, Image-to-Video, Reference-to-Video, and Video-to-Video workflows in ComfyUI, and claims improvements in HDR clarity, distant face restoration, motion fluidity, and fantasy VFX. The authors recommend pairing it with the minimax_h3_ref2v_turbo_4step_v0.1 LoRA for four-step accelerated inference, and an online demo is available via RunningHub.

Hugging Face trending models · 10d agoModel release7· 1 read

[AINews] Claude Fable/Mythos 5.1: new SOTA model, 75% cache price cut but 70% more output tokens

Anthropic launched Claude Fable 5.1 and Mythos 5.1, claiming new SOTA benchmarks, with 75% cache-read price cut and 1M-token context.

Anthropic released Claude Fable 5.1 and Mythos 5.1 as flagship models for coding and knowledge work, with a 1M-token context window and pricing of $10/$50 per million input/output tokens and cache reads cut 75% to $0.25. Artificial Analysis Intelligence Index scored Fable 5.1 at 66 versus 63 for Claude Opus 5, with HLE at 59.1% and Terminal-Bench v2.1 at 91.4%, though per-task cost rose ~20% due to 1.7x output token usage. Community analysis suggested Fable and Mythos may share underlying weights with different safety/routing behavior, and release notes highlighted Enterprise Frontier Safeguards and zero-data-retention support.

Latent Space · 14d agoModel release2

Retrospectively Reverse-Engineering Apple's Neural Engine

A developer reverse-engineers Apple's M1 Neural Engine architecture, mapping compute cores, MAC datapaths, and schedulers to explain the NPU's decline as transformers displaced CNN workloads.

A developer who previously maintained a reverse-engineered Linux driver for Apple's Neural Engine (ANE) published a retrospective deep dive mapping the M1 ANE's full internal architecture: compute, datapath, scheduler, memory, and execution model. The M1 ANE has 16 compute cores with 128 FP16 (or 256 INT8) MAC lanes each, totaling 2048 parallel MAC lanes, using 32-bit Q16.16 fixed-point accumulation with FP16 readout and an accumulator that saturates at 2^15. The author argues the ANE's dataflow was architected around the predictable reuse patterns of 2017-era CNN workloads (dating to the A11 Bionic), which autoregressive transformer decode broke, limiting its usefulness for general ML. With Apple's M5 folding ANE cores into GPU cores to tout LLM performance, the post frames this as the beginning of the end for the standalone NPU.

Google’s Gemini 3.8 Flash takes on bigger AI models at a lower cost

Google released Gemini 3.8 Flash with a security-focused Cyber variant that produces 2.6x more correct patches and found a critical vulnerability in under two hours.

Google launched Gemini 3.8 Flash for developers and a gated Gemini 3.8 Flash Cyber model reserved for vetted security teams through the new Fairwind program. The company says the model beats most larger frontier models on the DeepSWE v1.1 engineering benchmark at lower cost, and Chrome Security reports the Cyber variant produced 2.6 times more correct patches than the best commercial models while Google's Cloud Vulnerability Research team found a critical foundational vulnerability in under two hours. The models show significant prompt-injection robustness gains measured by Gray Swan and carry CBRN misuse safeguards, with cyber-offense restrictions on the standard version. Pricing matches Gemini 3.7 Flash at $0.75 per million input tokens and $3.75 per million output tokens.

Help Net Security · 13d agoModel release

Kaininja: Extending Native 3D Generators to the Part Level

KaiNinja extends TRELLIS.2 native 3D generation to part-level assets via a dual-volume O-Voxel representation, cutting whole-object Chamfer distance by 40%.

KaiNinja extends the TRELLIS.2 native 3D generator to produce part-level assets instead of one fused mesh, enabling downstream editing, rigging, and simulation. A dual-volume form of the O-Voxel representation solves the problem that a single volume cannot represent interfaces where two parts touch. The model needs no segmentation network, is partly trained on LLM-agent-authored part data, lowers whole-object Chamfer distance by 40%, and raises strict part F-score by 16% versus other part-generation pipelines.

Hugging Face daily papers · 2d agoAI research

unsloth/Qwen3.8-Flash-Next-GGUF — new model trending #21 on Hugging Face

Qwen released Qwen3.8-Flash-Next, an experimental 125B-parameter open-weight MoE previewing the Qwen4 architecture, with Unsloth shipping optimized GGUF quants.

Qwen released Qwen3.8-Flash-Next, an experimental open-weight preview of the architecture planned to underpin Qwen4. The model has 125B parameters with 6B activated, 512 experts (10 routed plus 1 shared), Qwen Sparse Attention (QSA), Gated DeltaNet, Gated Residual, and n-gram embeddings, with 262,144-token native context extendable to 1,000,000 tokens. Unsloth provides Dynamic 3.0 GGUF quantizations, and multi-token prediction (MTP) delivers 1.3-1.7x faster inference via llama.cpp or Unsloth Desktop.

Hugging Face trending models · 21d agoModel release1

Rosetta at AlexandriaX-2026: LoRA-Adapted NileChat for Context-Aware Dialectal Arabic Dialogue Translation

Rosetta ranks 4th and 5th in AlexandriaX-2026 dialectal Arabic dialogue translation using a LoRA adapter on NileChat-3B, finding limited pretraining benefit.

The Rosetta system for the AlexandriaX-2026 shared task fine-tunes a LoRA adapter on NileChat-3B for context-aware English-to-dialectal Arabic dialogue translation. The adapter was additionally pretrained on MADAR and PADIC dialect corpora for the unconstrained track. It achieved spBLEU 26.10 (4th, constrained) and 25.09 (5th, unconstrained). External dialect pretraining improved only two of thirteen dialects while slightly degrading overall performance, indicating negative transfer.

arXiv cs.AI / cs.LG / cs.CL · 6d agoAI research1

Real-World Knowledge-Guided Change Data Synthesis for Remote Sensing

Researchers introduce KnowChange, a framework that uses pretrained vision-language models to synthesize realistic change-detection training data for remote sensing.

KnowChange is a knowledge-guided change data synthesis framework that leverages pretrained vision-language models to reason about plausible change locations and class transitions from pre-change scenes and desired change types. It addresses the limited class-transition coverage and inflexibility of handcrafted rule-based synthesis methods, enabling diverse change types in a unified pipeline. Experiments show KnowChange-generated data outperforms existing synthetic datasets in both synthetic-to-real transfer and synthetic data augmentation, despite compact generation scale.

Hugging Face daily papers · 22d agoAI research

Domain-Specific Hallucination Detection in Large Language Models

A multi-signal pipeline detects LLM hallucinations, reaching F1 0.915 on HaluEval and cutting Qwen2.5-0.5B hallucination rates from 85.5% to 37.7% via DPO.

The paper presents a hallucination detection pipeline combining fine-tuned DeBERTa-v3 classification, Monte Carlo Dropout uncertainty, and temperature-scaled calibration. It achieves F1 0.915 and AUROC 0.977 on general-domain HaluEval tasks, with MC Dropout inference raising accuracy to 93.2%. Applying DPO to a Qwen2.5-0.5B generator reduces its hallucination rate from 85.5% to 37.7%, while cross-domain evaluation shows poor general-domain transfer to SciFact (F1 0.52) and PubMedBERT fine-tuning as the strongest adaptation (F1 0.63).

arXiv cs.AI / cs.LG / cs.CL · 5d agoAI research1

Marigold V2: Revisiting Diffusion Transformers for Monocular Depth Estimation

Marigold V2 adapts diffusion transformers for monocular depth estimation, improving AbsRel 16-26% over the previous best on KITTI and ETH3D.

Huawei's Bayer lab revisits the Marigold approach to repurpose image generation and editing models built on the diffusion transformer (DiT) architecture into monocular depth estimators. The recipes target single-step inference from pretrained multi-step flow-matching models, with remedies including alignment to ground-truth semantic features and a two-stage fine-tuning protocol using a Sinkhorn-based loss. The resulting model produces crisper depth maps that generalize out-of-distribution and also achieves state-of-the-art results on surface normals estimation and intrinsic image decomposition.

Hugging Face daily papers · 8d agoAI research

New AI Attack Hides Malicious Instructions in Normal-Looking Text to Evade Safety Filters

Check Point researchers show crafted prose hides policy-violating instructions that bypass all tested LLM gatekeepers, including GPT-4o mini and Llama Guard 3.

A new prompt-crafting technique embeds malicious payloads inside grammatical, natural-looking text without Base64, invisible Unicode, or obvious encodings, defeating lightweight pre-screening gatekeepers. In testing, all four evaluated gatekeeper models—gpt-4o-mini-2024-07-18, gpt-oss-safeguard:20b, claude-3-haiku-20240307, and llama-guard3:8b—classified the crafted wrappers as safe at a 100% bypass rate across 23 obfuscated prompts. GPT-5 Thinking in high-reasoning mode recovered and acted on the hidden instruction in 17 of 18 tests (~94.4%), often spending over a minute and multiple Python executions. Researchers recommend paraphrasing untrusted input, hardening gatekeeper policies, and applying defense-in-depth controls for agentic deployments.

GBHackers · 5d agoAI safety & security 2 sources

Adaptive Gated Deepfake Detection for Low-Resolution and Resource-Constrained Environments

AdaGate-DF routes deepfake detection by image quality through dual multi-exit gates, hitting 0.9370 AUC on Celeb-DF with low inference latency.

AdaGate-DF is an adaptive gated deepfake detection framework that uses image-quality cues to send high-quality images through earlier exits, saving compute in resource-constrained settings. On Celeb-DF it achieves an AUC of 0.9370, outperforming MaD-CoRN and DefakeHop++, and reaches 0.9708 at 384x384 resolution. On FaceForensics++ it remains effective under class imbalance while balancing uncertainty-aware prediction and computational efficiency.

arXiv cs.AI / cs.LG / cs.CL · 11d agoAI research

Safe word: What is it and why do you need one?

ESET recommends pre-agreed family safe words to counter AI voice-clone scams such as virtual kidnapping, as one-in-four Americans report receiving deepfake calls.

ESET outlines how scammers use just seconds of audio scraped from social media or work content to create convincing voice clones, with a Hiya report finding one-in-four Americans received a deepfake voice call in the past 12 months. Common schemes include virtual kidnapping calls mixing cloned voices with sobbing and background noise. A pre-agreed, non-OSINT-discoverable safe word, plus callback verification via known numbers and 2FA, reduces success rates of these frauds.

ESET WeLiveSecurity · 7d agoPhishing & fraud in the wild

Privacy Failure in Split-LLM Training, The Returned Gradient Nullifies the Decoys

Researchers show split-LLM training leaks privacy via zero-valued gradients on decoy rows, exposing which activations are real despite passing forward-channel checks.

A systems-security case study of a two-node split-LLM training setup found that the returned output gradient from an Untrusted Cloud Node is exactly zero for decoy rows, revealing which rows are real. Across nine seeds, zero patterns identified real rows in 4,096 of 4,096 frames per run, and an attack on frame contents recovered 0.65 to 1.50 percentage points of extra tokens over a baseline. Both datasets passed forward-channel privacy and quality checks but failed once the returned gradient was included. Row-wise gradient clipping and noise closed the leak for roughly 0.01 nats of held-out cross-entropy, though five unmeasured attack classes remain.

Training a 3.8B LLM to 0.384 CORE for $998 – Hugo Vergnes

Independent developer Hugo Vergnes trained a 3.8B-parameter Llama-style model to 0.384 CORE on 65B tokens for $998 in 43 hours on rented B200s.

Hugo Vergnes trained little-lm, a 3.848B-parameter decoder-only LLM, on 65.3B tokens in 43 hours for $998 using rented NVIDIA B200s, scoring 0.384 on the CORE benchmark and beating nanochat d32 (0.310) at similar cost. The Llama-style architecture uses RMSNorm, RoPE, GQA with 24 query and 8 KV heads, relu-squared MLPs, QK-norm, and ResFormer-style value embeddings that account for 19% of parameters. Key wins included the Muon optimizer for matrix parameters, a trapezoidal learning-rate schedule with linear cooldown, FP8 training plus vocabulary padding for roughly 33% throughput gains, and the ClimMix dataset over FineWeb-Edu. The project, inspired by Karpathy's nanochat, was built as a config-driven YAML framework for small LLM training.

New Ted Backdoor Hides Inside Victims' Own HAProxy Builds to Intercept Web Traffic

Rapid7 found a new backdoor, ted, compiled into trojanized HAProxy at two South Korean organizations, with medium-confidence attribution to North Korean actors.

Rapid7 documented a previously undocumented Linux toolkit named ted compiled into the HAProxy load balancer binaries of two South Korean organizations in the automotive and media sectors. The implant intercepts web traffic, serves altered pages only to filtered visitors, and hides C2 exchanges from backend logs and HAProxy statistics; a companion RAT, curlRAT, beacons on a default 12-hour schedule. The toolkit also trojanizes crond, sshd, agetty, atd, and polkitd binaries and sanitizes logs and bash history. Rapid7 attributes the activity with medium confidence to North Korean state-sponsored actors, with domain infrastructure overlapping APT37 listings in maltrail and delivery resembling the Operation SyncHole campaign.

The Hacker News · 11d agoThreat actor in the wild

Mind2Dialogue: Training Human-Aware Language Models by Simulating User Mental States

Mind2Dialogue simulates users' mental states to generate privileged supervision, boosting personalization and preference-following in Qwen, Llama, and OLMo assistants.

The Mind2Dialogue framework uses a psychology-guided simulator that preserves personal characteristics while updating user mental states through interaction, driving coherent conversations and an Oracle assistant's responses. Privileged distillation trains models on the Oracle's well-informed responses so they can assist users without direct access to mental states at deployment. Training on the full corpus improves every reported personalization metric over Qwen, Llama, and OLMo instruction-tuned baselines, including 26.6 to 40.9 percentage point gains in preference-following generation.

Hugging Face daily papersupdated · 1d agofirst · 2d agoAI research 2 sources