LoRA, PEFT, instruction tuning and domain adaptation — adapting a pretrained model without paying to train one.
arXiv:2608.30795v1 Announce Type: cross Abstract: End-to-end weather forecasting systems produce skillful global gridded and station forecasts directly from raw Earth observations, replacing the nume...
The paper introduces a German-English benchmark dataset to evaluate anti‑LGBTQ biases in language models, combining community‑sourced stereotypes from German‑speaking queer individuals with a German translation of WinoQueer. Eight language models of varying sizes and architectures were assessed, revealing that they reproduce anti‑queer stereotypes with differences across identities and models. Fine‑tuning on community and progressive media content reduced bias on average, though the effect was not consistent across all models and identities.
PaperGym is a framework that transforms each research paper into a training environment for AI research planning, using rubrics extracted from the paper’s method and experiments as a critic. It synthesizes research questions from the goal and background, and derives evaluation criteria from the method and experiments, reducing criterion leakage to 3.7%. Experiments with Qwen models show that training with PaperGym’s rubric improves benchmark performance by up to 5.6 points and outperforms existing datasets and fine‑tuning baselines.
arXiv:2608.29461v1 Announce Type: new Abstract: Sandbagging models strategically underperform on evaluations while retaining the capabilities being measured. The evaluations that guide frontier-model...
Padàrtha is the first ontology‑grounded fine‑grained Named Entity Recognition benchmark for Classical Sanskrit, built on the Mahêbhárata epic. Its tag set, derived from the Nyáya‑Vai’séka ontological system, contains 18 fine‑grained categories under 10 nodes and maps to five standard coarse tags, ensuring compatibility with existing benchmarks. The dataset includes over 12.6K expert‑annotated entries and 108,335 entity mentions across 73,632 verses, plus a 5,000‑verse test set designed to challenge rare mentions, and the study compares generative NER models to traditional architectures, noting performance drops at finer granularity and difficulties with unseen entities.
The paper introduces Engram Adapter, a conditional memory adapter that selectively activates on in-domain inputs while suppressing out-of-domain (OOD) retrievals. It leverages multi‑channel matching over local n‑gram patterns with occupancy tracking and a learned scalar gate to inject residuals only when appropriate. Experiments on Qwen3-4B/8B with AG‑News and MedMCQA show improved in‑domain accuracy while preserving nearly all OOD performance, outperforming always‑on baselines on LegalBench.
The paper introduces Juris Policy Optimization (JPO), a post‑training framework designed to enhance structured legal reasoning in Chinese criminal judgment prediction. JPO first trains models with teacher‑generated rationales to guide a four‑step reasoning process, then applies reinforcement learning using a composite reward that balances prediction accuracy, reasoning completeness, and cross‑step consistency. Experiments on several open‑source language models and three Chinese legal benchmarks demonstrate that JPO consistently outperforms both supervised fine‑tuning and standard reinforcement learning baselines in terms of judgment prediction and reasoning quality.
The paper reports that chain‑of‑thought (CoT) supervised fine‑tuning (SFT) improves reasoning but systematically harms long‑context recall in hybrid linear‑attention models such as HypeNet and Jet‑Nemotron. Retrieval performance on the Needle‑In‑A‑Haystack benchmark drops dramatically after CoT‑SFT, especially with harder settings and longer contexts. The authors introduce QK‑Restore, a training‑free method that reinstates the query‑key projection matrices from the pre‑SFT checkpoint, which recovers long‑range recall while preserving reasoning gains.
The paper introduces a diagnostic framework to disentangle the effects of character profile axes—Familiarity, Structure, and Disposition—on large language model role‑playing agents. Experiments on 211 personas and five LLMs show that Familiarity and Structure have little impact, whereas Disposition, particularly immoral traits, consistently degrades performance. The authors propose Field‑Aware Contrastive Decoding (FACD), a training‑free method that mitigates this performance gap without harming moral‑character performance.
arXiv:2608.29948v1 Announce Type: new Abstract: Evaluating data-text alignment remains challenging: existing metrics often provide limited explanations for the scores, while prompt-based LLM-as-Judge...
arXiv:2608.30325v1 Announce Type: new Abstract: Natural-language instructions enable flexible control of synthesized speech, yet emotional TTS systems primarily model a single utterance-level affect,...
arXiv:2608.30426v1 Announce Type: new Abstract: Current dialogue systems struggle with dynamic information retrieval, often leading to hallucinations and lower response accuracy. We address this by a...
Randomized YaRN is a training method that enhances length generalization for large language models by combining YaRN-based positional extrapolation with randomized positional encoding and a length curriculum. During training on short-context data, tokens receive YaRN positional encodings sampled from a larger position range, exposing the model to out-of-distribution positional representations. Evaluated on BABILong, Multi-Round Coreference Resolution, and LongBench v2, Randomized YaRN consistently improves reasoning performance on context lengths from 16K to 128K, outperforming standard fine‑tuning especially at far out‑of‑distribution lengths.
Graph2Counsel is a framework that generates synthetic counseling dialogues by leveraging Client Psychological Graphs (CPGs) to encode the relationships among a client’s thoughts, emotions, and behaviors. The system uses a structured prompting pipeline guided by counselor strategies and explores techniques such as Chain‑of‑Thought and Multi‑Agent Feedback to produce 760 realistic sessions from 76 CPGs. Expert evaluation shows the dataset surpasses previous ones in specificity, counselor competence, authenticity, conversational flow, and safety, and fine‑tuning an open‑source model on it improves performance on several counseling benchmarks.
arXiv:2608.28675v1 Announce Type: cross Abstract: Video reasoning tasks such as grounded video question answering and temporal grounding require selecting temporal evidence that supports the query. I...
arXiv:2505.22635v2 Announce Type: replace-cross Abstract: A common approach for teaching large language models (LLMs) to reason is to train on chain-of-thought (CoT) traces of in-distribution reasoni...
arXiv:2608.29680v1 Announce Type: new Abstract: Feed-forward 3D foundation models reconstruct perspective scenes in one pass. Satellite photogrammetry needs a different product, one that domain adapt...
arXiv:2608.29186v1 Announce Type: new Abstract: Federated Video Domain Adaptation (FVDA) enables collaborative learning across distributed and non-IID video datasets while preserving privacy, but is...
arXiv:2608.29733v1 Announce Type: new Abstract: Visual aliasing, also known as the doppelganger problem, remains a key challenge for structure-from-motion (SfM): visually similar but physically disti...
arXiv:2602.01605v2 Announce Type: replace Abstract: Time Series Foundation Models (TSFMs) leverage extensive pretraining to accurately predict unseen time series during inference, without the need fo...