HinTel-AlignBench: A Framework and Benchmark for Hindi-Telugu with English-Aligned Samples

The SAS Academy for Data & AI Excellence is actively training India's workforce to adapt to the emerging Generative AI (GenAI) era. This initiative aims to equip professionals with the necessary skills to thrive in a rapidly evolving technological landscape, thereby enhancing India's position in the global AI market.

The article discusses EyeVLA, a robotic eyeball designed for active visual perception in embodied AI systems. Unlike traditional models that passively process images, EyeVLA actively acquires detailed information while managing spatial constraints. This innovation aims to enhance the effectiveness of robotic applications in open-world environments by integrating action tokens with vision-language models (VLMs) for improved understanding and interaction.

Large Language Models (LLMs) have shown exceptional ability in text generation but often produce factually incorrect statements, known as 'hallucinations'. This study investigates hallucinations in conversational data across three low-resource languages: Hindi, Farsi, and Mandarin. The analysis of various LLMs, including GPT-3.5 and GPT-4o, reveals that while Mandarin has few hallucinated responses, Hindi and Farsi exhibit significantly higher rates of inaccuracies.

The paper discusses TeamNRC's performance in the BHASHA-Task 1 Grammatical Error Correction shared task, focusing on five Indian languages. The approach utilized zero/few-shot prompting of various language models, achieving notable ranks in Telugu and Hindi with GLEU scores of 83.78 and 84.31, respectively. The study also examines data quality and evaluation metrics, emphasizing the potential of smaller language models for Indian scripts.

The paper presents a new metric called Physics-Constrained Multimodal Data Evaluation (PCMDE) aimed at improving the evaluation of multimodal synthetic images. Current metrics like BLEU and CIDEr often fail to accurately assess semantic and structural accuracy, particularly in specific domains. PCMDE integrates large language models with reasoning and vision-language models to enhance feature extraction, validation, and physics-guided reasoning.

The article discusses the evaluation of video models' reasoning abilities through maze-solving tasks, introducing VR-Bench, a benchmark with 7,920 procedurally generated videos. This research aims to determine if video models can reason via video generation, leveraging spatial layouts and temporal continuity for effective reasoning. The findings indicate that SFT can effectively elicit reasoning capabilities in video models.

India's technology minister announced that the country's chipmaking capabilities are expected to match those of major global producers by 2032. This ambitious timeline reflects the government's commitment to enhancing domestic manufacturing in the technology sector.

India generates 20% of global data but stores only 3% of it locally. This discrepancy is being addressed by two significant forces that aim to enhance the country's data storage capabilities. The ongoing data centre boom in India is expected to bridge this gap, facilitating better data management and storage solutions.