arXiv subjects

Saurav Bhandari

Publications and source records attributed to Saurav Bhandari.

3 recordsLinked to original sources

arXiv2026

Steering Under Compression: Dose-Response, Capability Cost, and Failure Asymmetry in Quantized LLMs

Inference-time activation steering enables behavioral control of large language models without parameter modification, while post-training quantization reduces memory and compute costs for deployment. Despite their growing convergence in practice, the interaction between these two techniques remains uncharacterized. We systematically study activation steering under weight-only quantization (INT8 and NF4) across four open-weight 7-9B models and two behavioral targets: judged sentiment and judge-free reasoning length. Using an iso-effect framework that compares capability costs at matched behavioral effect, we find that sentiment steering survives quantization intact. After correcting a GSM8K parser artifact with a uniform v2.3.1 rescore, the pooled INT8 contrast is -0.010 (90% CI [-0.026, +0.007]), descriptively Equivalent under the preregistered three-label rule, while NF4 remains Inconclusive at -0.017 ([-0.067, +0.033]). In contrast, reasoning length exhibits a surprising asymmetric dose-response: lengthening is graded but terminates in cap-runaway and collapse, while shortening is a step function with only 12-30% shortening (model-dependent) before discontinuous failure. We expose a methodological pitfall: the naive iso-effect ladder anchors on the collapse floor for floor-bounded targets, and we introduce a censored construction that restores interpretable crossings. We also quantify a substantial baseline capability shift for Mistral-NF4 (0.545 to 0.365 GSM8K at alpha=0), demonstrating that compression can dominate the steering intervention. Despite this, steering vectors remain highly collinear with their FP16 siblings (cosine similarity 0.989-0.998 for INT8, 0.945-0.990 for NF4), confirming that the behavioral direction survives quantization even when the cost structure does not. All code and data are released.

cs.LG↗

arXiv2025

Nepali Sign Language Characters Recognition: Dataset Development and Deep Learning Approaches

Sign languages serve as essential communication systems for individuals with hearing and speech impairments. However, digital linguistic dataset resources for underrepresented sign languages, such as Nepali Sign Language (NSL), remain scarce. This study introduces the first benchmark dataset for NSL, consisting of 36 gesture classes with 1,500 samples per class, designed to capture the structural and visual features of the language. To evaluate recognition performance, we fine-tuned MobileNetV2 and ResNet50 architectures on the dataset, achieving classification accuracies of 90.45% and 88.78%, respectively. These findings demonstrate the effectiveness of convolutional neural networks in sign recognition tasks, particularly within low-resource settings. To the best of our knowledge, this work represents the first systematic effort to construct a benchmark dataset and assess deep learning approaches for NSL recognition, highlighting the potential of transfer learning and fine-tuning for advancing research in underexplored sign languages.

cs.CV↗

arXiv2025

Diversity Conscious Refined Random Forest

Random Forest (RF) is a widely used ensemble learning technique known for its robust classification performance across diverse domains. However, it often relies on hundreds of trees and all input features, leading to high inference cost and model redundancy. In this work, our goal is to grow trees dynamically only on informative features and then enforce maximal diversity by clustering and retaining uncorrelated trees. Therefore, we propose a Refined Random Forest Classifier that iteratively refines itself by first removing the least informative features and then analytically determines how many new trees should be grown, followed by correlation-based clustering to remove redundant trees. The classification accuracy of our model was compared against the standard RF on the same number of trees. Experiments on 8 multiple benchmark datasets, including binary and multiclass datasets, demonstrate that the proposed model achieves improved accuracy compared to standard RF.

cs.LG↗