The llama.cpp release b9661 adds GGML_OP_COL2IM_1D support for Vulkan, using a bounded gather loop instead of full-K scan with modulo. It returns nullptr for unsupported types and includes builds for macOS, Linux, Android, Windows, and openEuler across CPU, Vulkan, CUDA, and SYCL.
Claude Fable 5 was banned under export controls after researchers demonstrated it could 'fix' code with known vulnerabilities. The model successfully generated patches and test scripts for security flaws, a capability essential for defensive cybersecurity. The researchers argue this is a legitimate security function, not a threat, and that banning such models undermines real-world cyber defense.
Users have asked whether running multiple machines in parallel provides advantages for larger context handling or faster inference in local large language models. While individual machines can handle larger contexts with sufficient RAM, there is no established advancement enabling significant performance gains from distributing inference across multiple machines for local LLMs.
Users report inconsistent results when using quantized models in image generation, with SD 1.5 working well but SDXL failing. Despite successful conversion and quantization using tools like convert.py and llama-quantize, some users obtain poor outputs while others do not, raising questions about the current state and reliability of quantized image generation technology.
The Nex2 mini Phase Twin, a 30B parameter model with 16GB footprint, is now available for Intel users, particularly the A770 lineup. It performs at 89 tokens per second on a single A770 card and is optimized to use the appropriate kernel based on hardware, with enhanced performance when paired with two cards.
The DGX Spark is being unfairly criticized despite its strong scalability and usable local AI performance. Its ConnectX technology allows lossless expansion, and at 240W power, it enables running agentic DS4Flash locally for around $9k with 256GB of CUDA memory.
Katie Moussouris, a cybersecurity expert, reported that Anthropic shared the White House's Fable jailbreak report with her for evaluation. She noted that Fable refused to analyze insecure code but complied when asked to fix it, describing this as the model functioning as intended in cyberdefense.
A training-free diagnostic, contrastive-difference CKA (CKA_Delta), identifies concept-specific structural alignment across language model architectures. It detects geometric convergence and functional transfer across six concept domains, including non-instructional tasks, with significant discrimination where standard CKA fails. Results suggest universality may strengthen with model scale, though further validation is needed.
The Informath project demonstrates symbolic informalization to convert formal mathematical proofs into fluent, precise natural language. It uses Dedukti as a hub connecting proof systems like Agda, Lean, and Rocq, with Grammatical Framework ensuring linguistic correctness across multiple languages.
LOGOS is a unified generative language model that represents scientific objects and their interactions as token sequences in a shared grammar. It achieves consistent or superior performance across diverse natural science tasks, demonstrating the feasibility of a single model serving multiple domains. The model scales positively with parameter count, and its design suggests that AI for Science should align deeply with large language models through shared architectures and training.
LESS introduces a training-free, model-agnostic adaptive sampler that reduces reverse denoising steps by 72.1% compared to fixed-budget decoding. It achieves higher accuracy than existing training-free samplers and lowers inference compute and latency through mutual-stability rules that ensure token commitment only when predictions are confident, consistent, and stable.
IMPACTeen is a dataset of 1,021 texts annotated from five perspectives—teenagers, parents, psychologists, communication experts, and teachers. It includes 5,100 annotation records covering social influence techniques, intentions, consequences, and resistance, with annotations validated through human editing. The dataset, created using LLM generation and human validation, is available in both Polish and English and supports research on social influence and language model training.
A study identifies extrinsic (crucial tokens) and intrinsic (cognitive behaviors) properties that enhance code interpreter reasoning in large language models. Stronger reasoning models show higher prevalence of verification, backtracking, and backward chaining, with these properties improving performance during inference and training, reducing overthinking and boosting token efficiency.
A measurement study finds that 26 semantic post-hoc operators do not improve held-out accuracy over Best-of-N in frozen small code models. While two operators—expression-layer recovery and adaptive consensus early-stop—offer benefits in compute efficiency or program recovery, none outperform BoN in accuracy. The results highlight systemic limitations in error detection and coverage, suggesting that model harnesses and error coverage must be improved before post-hoc reasoning is considered.
TokenPilot reduces inference costs by 61% to 87% in both isolated and continuous modes, outperforming prior systems in cost efficiency while maintaining competitive performance. It uses ingestion-aware compaction and lifecycle-aware eviction to preserve prompt cache continuity and minimize token footprints.
DeepRubric introduces a data construction framework that builds query-rubric pairs by first defining verifiable evaluation targets through an evidence tree. It generates 9K supervision examples and trains a 8B model with GRPO, achieving performance comparable to state-of-the-art models using 13x fewer RL GPU-hours.
KVEraser enables efficient localized context erasing in large language models by replacing only the KV cache states of an erased span with learned steering states. It achieves near-full-recomputation performance on in-domain tasks across 1K to 32K context lengths, with only a 24% latency increase, and outperforms other approximate methods in long-document QA with 3--4x speedup over full recomputation.
MetaSyn introduces a dataset of 442 expert-curated meta-analyses from Nature Portfolio. It evaluates twelve LLM agent configurations and reveals a critical bottleneck in study screening, where no system recovers more than 52.7% of ground-truth included literature despite high retrieval recall.
ContextRL introduces an indirect auxiliary objective to improve long-horizon reasoning and multimodal performance in LLMs. It rewards models for selecting the context that supports a query-answer pair, using contrastive context data from coding agent trajectories and image-based visual questions. ContextRL achieves +2.2% and +1.8% gains over standard methods on long-horizon and visual QA benchmarks, with gains attributed to the selection objective, not data augmentation.
Qwen3-8B internally tracks the value of its current trajectory, defined as the likelihood of achieving its goals. This 'value' axis distinguishes confidence levels, backtracking behavior, and code correctness, and shows that preference optimization boosts confidence in rewarded behaviors. The model assigns low value to politically sensitive queries post-training, and fine-tuning increases confidence within specific domains.