Large language model (LLM) news — GPT-4, Claude, Gemini, Llama, Mistral and the latest research on training, fine-tuning, RLHF, and deployment of LLMs.
A study of 93 Urdu stories generated by three contemporary LLMs — GPT-5.1, Qwen-3-Max, and DeepSeek-3.1 — found that the models frequently make basic grammar and semantic errors, produce incoherent and repetitively unnat…
A prompt-based approach to minimal-edit grammatical error correction reached an F0.5 score of 78.32 on the BEA-2019 test set, closing the gap to the fine-tuned single-model state of the art (Staruch et al., 2025) to 0.38…
A new arXiv paper (2609.10792v1) tests whether transformer-based models internalize the same linguistic features as traditional feature-based classifiers in Automatic Readability Assessment across Arabic, English, French…
A new arXiv paper (2609.10830v1) reports that language models' apparent memory of training data is largely confounded, using exact duplication counts from the published pretraining corpora of OLMo-2 and Pythia. Across fi…
A new arXiv paper (2609.10934v1) reports that modeling semantic uncertainty — an LLM-derived measure of how strongly a turn so far constrains what may plausibly come next — substantially outperforms prompt-based and fine…
Researchers introduced LLM-Anchored Paralinguistic Enrichment (LAPE), a method that combines LLM-derived linguistic representations with paralinguistic speech cues to detect Alzheimer's disease, achieving state-of-the-ar…
A new arXiv paper introduces SearchAtlas, a framework that converts LLM search-agent trajectories into structured graphs whose edges trace how evidence propagates from query to final answer. The automated parsing pipelin…
A new arXiv paper (2609.10923v1) introduces "Structurally Speaking," a lightweight structured prompting protocol that guides large language models to translate between explicit graph connectivity and motif-level abstract…
A new arXiv paper (2609.10996v1) reports that verbalized confidence has become the more robust soft-scoring mechanism for LLM-as-a-Judge on post-2025 proprietary models, reversing the standard advice to prefer log-probab…
A study of five realistic noise conditions applied to 3,822 stereotype-related responses found that surface noise in text — typos, informal spelling, and broken punctuation — makes large language model judges far more li…
Researchers released OpenDiscoveryTrace, a public dataset of 558 complete AI scientific agent trajectories across 124 scientific tasks in drug discovery, materials science, genomics, and literature analysis, spanning sev…
A new arXiv paper (2609.09233v1) reports that executing reusable agent skill packages as subagents outperforms loading skill instructions directly into a main agent's context window on long-horizon tasks. The authors fin…
Researchers introduced GeoSteer, an optimization-based method for norm-preserving activation steering in large language models that formulates steering as a Riemannian optimization problem and updates activations through…
A new arXiv paper (2609.10863v1) identifies a duality between continuous and discrete flow matching, showing that projecting continuous convex-interpolant paths with one-hot targets through a position-wise argmax yields …
Researchers introduced MHE-Former, a Transformer-based multi-hypothesis framework that uses entropy maximization to generate diverse 3D hand and body mesh recovery predictions from monocular input, according to a paper p…
A Kubernetes Dynamic Resource Allocation (DRA) driver makes composable CXL memory a schedulable cluster resource for cross-node KV-cache reuse in LLM serving, according to an arXiv paper (2609.10790v1). On a two-node clu…
A production skill router covering 34,396 skills shows that fine-tuning on synthetic data improves in-distribution skill retrieval for LLM agents but causes catastrophic forgetting on real and out-of-distribution data, a…
A new arXiv paper (2609.10613v1) proposes a posterior reweighting framework that models safety-aligned multimodal large language models (MLLMs) as implicitly operating over competing behavioral modes, explaining in-conte…
Researchers introduced SolCloudLLM, an LLM-based multimodal forecasting framework that fuses ground-based sky images with historical time-series data through bidirectional multimodal fusion to predict short-term photovol…
A new arXiv paper (2609.11699v1) introduces Negative Self-Distillation (NSD), a framework that improves large language model reasoning by diverging from self-generated flawed reasoning rather than imitating privileged gr…