arXiv subjects

Amit Garg

Publications and source records attributed to Amit Garg.

5 recordsLinked to original sources

arXiv2025

Phi-4-Mini Technical Report: Compact yet Powerful Multimodal Language Models via Mixture-of-LoRAs

We introduce Phi-4-Mini and Phi-4-Multimodal, compact yet highly capable language and multimodal models. Phi-4-Mini is a 3.8-billion-parameter language model trained on high-quality web and synthetic data, significantly outperforming recent open-source models of similar size and matching the performance of models twice its size on math and coding tasks requiring complex reasoning. This achievement is driven by a carefully curated synthetic data recipe emphasizing high-quality math and coding datasets. Compared to its predecessor, Phi-3.5-Mini, Phi-4-Mini features an expanded vocabulary size of 200K tokens to better support multilingual applications, as well as group query attention for more efficient long-sequence generation. Phi-4-Multimodal is a multimodal model that integrates text, vision, and speech/audio input modalities into a single model. Its novel modality extension approach leverages LoRA adapters and modality-specific routers to allow multiple inference modes combining various modalities without interference. For example, it now ranks first in the OpenASR leaderboard to date, although the LoRA component of the speech/audio modality has just 460 million parameters. Phi-4-Multimodal supports scenarios involving (vision + language), (vision + speech), and (speech/audio) inputs, outperforming larger vision-language and speech-language models on a wide range of tasks. Additionally, we experiment to further train Phi-4-Mini to enhance its reasoning capabilities. Despite its compact 3.8-billion-parameter size, this experimental version achieves reasoning performance on par with or surpassing significantly larger models, including DeepSeek-R1-Distill-Qwen-7B and DeepSeek-R1-Distill-Llama-8B.

cs.CL↗

arXiv2024

Phi-3 Safety Post-Training: Aligning Language Models with a "Break-Fix" Cycle

Recent innovations in language model training have demonstrated that it is possible to create highly performant models that are small enough to run on a smartphone. As these models are deployed in an increasing number of domains, it is critical to ensure that they are aligned with human preferences and safety considerations. In this report, we present our methodology for safety aligning the Phi-3 series of language models. We utilized a "break-fix" cycle, performing multiple rounds of dataset curation, safety post-training, benchmarking, red teaming, and vulnerability identification to cover a variety of harm areas in both single and multi-turn scenarios. Our results indicate that this approach iteratively improved the performance of the Phi-3 models across a wide range of responsible AI benchmarks. Finally, we include additional red teaming strategies and evaluations that were used to test the safety behavior of Phi-3.5-mini and Phi-3.5-MoE, which were optimized for multilingual capabilities.

cs.CL↗

arXiv2024

Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone

We introduce phi-3-mini, a 3.8 billion parameter language model trained on 3.3 trillion tokens, whose overall performance, as measured by both academic benchmarks and internal testing, rivals that of models such as Mixtral 8x7B and GPT-3.5 (e.g., phi-3-mini achieves 69% on MMLU and 8.38 on MT-bench), despite being small enough to be deployed on a phone. Our training dataset is a scaled-up version of the one used for phi-2, composed of heavily filtered publicly available web data and synthetic data. The model is also further aligned for robustness, safety, and chat format. We also provide parameter-scaling results with a 7B, 14B models trained for 4.8T tokens, called phi-3-small, phi-3-medium, both significantly more capable than phi-3-mini (e.g., respectively 75%, 78% on MMLU, and 8.7, 8.9 on MT-bench). To enhance multilingual, multimodal, and long-context capabilities, we introduce three models in the phi-3.5 series: phi-3.5-mini, phi-3.5-MoE, and phi-3.5-Vision. The phi-3.5-MoE, a 16 x 3.8B MoE model with 6.6 billion active parameters, achieves superior performance in language reasoning, math, and code tasks compared to other open-source models of similar scale, such as Llama 3.1 and the Mixtral series, and on par with Gemini-1.5-Flash and GPT-4o-mini. Meanwhile, phi-3.5-Vision, a 4.2 billion parameter model derived from phi-3.5-mini, excels in reasoning tasks and is adept at handling both single-image and text prompts, as well as multi-image and text prompts.

cs.CL↗

arXiv2022

Quantifiable Assurance: From IPs to Platforms

Hardware vulnerabilities are generally considered more difficult to fix than software ones because they are persistent after fabrication. Thus, it is crucial to assess the security and fix the vulnerabilities at earlier design phases, such as Register Transfer Level (RTL) and gate level. The focus of the existing security assessment techniques is mainly twofold. First, they check the security of Intellectual Property (IP) blocks separately. Second, they aim to assess the security against individual threats considering the threats are orthogonal. We argue that IP-level security assessment is not sufficient. Eventually, the IPs are placed in a platform, such as a system-on-chip (SoC), where each IP is surrounded by other IPs connected through glue logic and shared/private buses. Hence, we must develop a methodology to assess the platform-level security by considering both the IP-level security and the impact of the additional parameters introduced during platform integration. Another important factor to consider is that the threats are not always orthogonal. Improving security against one threat may affect the security against other threats. Hence, to build a secure platform, we must first answer the following questions: What additional parameters are introduced during the platform integration? How do we define and characterize the impact of these parameters on security? How do the mitigation techniques of one threat impact others? This paper aims to answer these important questions and proposes techniques for quantifiable assurance by quantitatively estimating and measuring the security of a platform at the pre-silicon stages. We also touch upon the term security optimization and present the challenges for future research directions.

cs.CR↗

arXiv2015

Exploring Correlation between Labels to improve Multi-Label Classification

This paper attempts multi-label classification by extending the idea of independent binary classification models for each output label, and exploring how the inherent correlation between output labels can be used to improve predictions. Logistic Regression, Naive Bayes, Random Forest, and SVM models were constructed, with SVM giving the best results: an improvement of 12.9\% over binary models was achieved for hold out cross validation by augmenting with pairwise correlation probabilities of the labels.

cs.LG↗