LoRA, PEFT, instruction tuning and domain adaptation — adapting a pretrained model without paying to train one.
arXiv:2604.11788v2 Announce Type: replace Abstract: High dynamic range (HDR) imagery offers a rich and faithful representation of scene radiance, but remains challenging for generative models due to...
The paper introduces Adjacent Possible Exploration (APE), a selective fine‑tuning strategy for large language models that evaluates multiple parameter updates on small data subsets and accepts only those that surpass a performance threshold. APE draws on evolutionary optimization ideas to filter out destabilizing changes, contrasting with conventional single‑gradient fine‑tuning. Experiments on news summarization show a 33.9% BLEU gain and a 36.2% perplexity drop while keeping computational demands low.
The paper introduces phoneme‑guided initialization for large language model‑based speech recognition. By pre‑training an audio encoder on speech‑to‑phoneme conversion and an LLM on phoneme‑to‑grapheme conversion, then fine‑tuning jointly on ASR, the method leverages phoneme‑mediated processing within an end‑to‑end framework. Experiments on Japanese, Chinese, Tatar, and Urdu demonstrate that this approach matches or surpasses both cascaded S2P‑P2G baselines and end‑to‑end models without phoneme initialization.
The paper examines whether word error rate adequately measures privacy in speech by applying entity‑aware metrics from NLP to evaluate two obfuscation techniques. It focuses on protecting named entities and explores how fine‑tuning on entity‑rich data affects attack success across different entity categories. The authors also offer guidance on choosing metrics depending on whether the obfuscation preserves temporal alignment.
arXiv:2610.08810v1 Announce Type: new Abstract: Fine-tuned geospatial foundation models (GeoFMs) pretrained on large satellite archives have been shown to improve crop classification accuracy and geo...
The paper identifies a new failure mode in Video Large Language Models (Video‑LLMs) called directional motion blindness, where models struggle to correctly interpret basic signed image‑plane motion directions. By tracing motion signals through the model pipeline, the authors reveal a binding gap between motion information and verbal output, and propose MoDirect, a motion‑direction instruction‑tuning dataset, and DeltaDirect, a projector‑level objective that predicts 2‑D motion vectors from frame deltas. Experiments show that DeltaDirect dramatically boosts motion‑direction accuracy on both synthetic and real‑world benchmarks while maintaining overall video‑understanding performance.
The paper examines the vulnerability of open‑weight large language model safeguards that aim to block harmful usage by assuming new harmful behavior arises only through fine‑tuning. It demonstrates that simple, low‑cost attacks—abliteration and prefilling—can significantly increase attack success rates across multiple benchmarks, and proposes an abliteration‑resistant tuning (ART) method that mitigates these attacks. The study shows that safeguarding defenses must be evaluated against a broader range of attack strategies beyond fine‑tuning.
The paper investigates whether removing harmful information from open‑weight models can guarantee resistance to fine‑tuning attacks. It demonstrates that mutual information measured at model release is insufficient to certify tamper resistance, because function‑preserving reparameterizations can keep information unchanged while altering the gradient‑descent dynamics. The authors show that training order and parameterization can significantly affect recovery time, even when both weight–data and label–representation mutual information are zero, and they propose that certification must include constraints on attack dynamics beyond mutual information.
The paper argues that pattern recognition and step‑by‑step reasoning lie on a spectrum, with large language models (LLMs) learning the latter when the next token depends on only a few preceding tokens. It formalises reasoning traces as paths on a De Bruijn graph, showing that the number of edges is far smaller than the number of possible traces, making step‑by‑step reasoning sample‑efficient. Experiments fine‑tuning Qwen2.5‑1.5B‑Instruct demonstrate that a moderate density of states balances accuracy and robustness, and that real‑world models like Qwen3 retain most of their performance even when attention is limited to a small sliding window.
The paper presents the Itgan systems for the NADI 2026 Arabic ASR shared task, covering robust country-level, mixed-dialect, and Tunisian code‑switched subtasks. All systems use a Whisper model adapted with LoRA on consumer GPUs, with task‑specific additions: per‑dialect specialists for 1.1, base‑model choice and decorrelated ensemble for 1.2, and a weight‑space average plus ROVER voting for 1.3. The submissions achieved 57.1% WER on 1.1, 46.7% WER on 1.2, and 14.49% WER on 1.3, with detailed statistical comparisons and noted ineffective approaches.
EmphTTS is a non‑autoregressive text‑to‑speech system that uses Group Relative Policy Optimization (GRPO) to train its duration predictor with an emphasis localization reward, allowing direct word‑level emphasis control. The system outperforms existing baselines in both objective emphasis evaluation and subjective preference tests, and ablation studies confirm that GRPO improves emphasis realization while reducing mismatch between the duration predictor and the TTS model.
arXiv:2610.09109v1 Announce Type: new Abstract: Many organizations fine-tune publicly available pretrained Automatic Speech Recognition (ASR) models and deploy them in black-box settings, assuming li...
arXiv:2610.09549v1 Announce Type: new Abstract: And-Inverter Graphs (AIGs) are fundamental representations for logic synthesis and verification in Electronic Design Automation (EDA). As structured re...
arXiv:2610.10203v1 Announce Type: new Abstract: Model organisms of alignment-relevant behaviors (e.g., backdoors, sycophancy, spurious correlations) have emerged as a key tool for evaluating whitebox...
arXiv:2610.09384v1 Announce Type: cross Abstract: Language models are routinely fine-tuned under a fixed context, such as a generic system prompt, persona or domain-specific instruction, yet the lear...
arXiv:2610.09819v1 Announce Type: cross Abstract: Acoustic foundation models (AFMs) have democratized acoustic applications, enabling powerful models for tasks ranging from speech recognition to spea...
arXiv:2610.09943v1 Announce Type: cross Abstract: Reinforcement learning (RL) fine-tuning improves vision-language-action (VLA) policies through closed-loop experience, yet generalization beyond the...
arXiv:2610.10261v1 Announce Type: cross Abstract: Context: Once defined a taxonomy of stages structuring Machine Learning (ML) pipelines (e.g. Data Preprocessing, Modeling...), extracting these stage...
arXiv:2610.10407v1 Announce Type: cross Abstract: As option markets grow and AI advances, agentic systems for option trading are gaining increasing attention. Language-model-based agents can reason o...
arXiv:2605.14605v3 Announce Type: replace-cross Abstract: Model providers increasingly release the weights of large language models. Although these models are safety-aligned before release, their saf...