arXiv subjects

Shantanu Godbole

Publications and source records attributed to Shantanu Godbole.

7 recordsLinked to original sources

arXiv2026

Depth-Aware Sensitivity Analysis of Mixture-of-Experts Models via Magnitude-Based Expert Masking

Mixture-of-Experts (MoE) architectures scale large language models (LLMs) while preserving computational efficiency through sparse activation. Despite their widespread adoption, the relative importance of individual MoE layers remains insufficiently characterized, particularly for model compression. This paper presents a systematic layer-wise sensitivity analysis of the Qwen3.6-35B-A3B model (40 MoE layers, 256 experts per layer, top-8 routing) using magnitude-based expert masking on the XLCoST cross-lingual code translation benchmark. We conduct a multi-phase study spanning 100, 300, and 500 prompt evaluation scales across three H100 GPU servers. Our central finding is that layer sensitivity is strongly depth-dependent: early layers (0-9) and middle layers (10-29) are highly fragile to expert masking, while late layers (30-39), and especially very-late layers (35-39), tolerate aggressive masking of low-magnitude experts. Flat all-layer masking at 30% retains only 150/300 Good+Similar outputs at 300-prompt scale, whereas late-focused policies retain 249-255/300 while masking 640-1,145 experts. On a later 500-prompt held-out validation slice, the narrow very-late policy (layers 35-39 @ 50%) achieves the strongest quality/masked-expert tradeoff among tested candidates, retaining 419/500 Good+Similar outputs while masking only 640 of 10,240 total experts. We additionally characterize top-k routing width reduction from 8 to 6 active experts per token, which shows a large observed wall-clock reduction on a 100-prompt probe with no Good+Similar loss, though it does not yet compose cleanly with aggressive expert masking. These findings provide an empirical foundation for depth-aware MoE expert masking and establish a practical path toward physical weight surgery, activation-based expert scoring, and training-based recovery.

cs.AI↗

arXiv2026

Attention Mechanism and Heuristic Approach: Context-Aware File Ranking Using Multi-Head Self-Attention

The identification and ranking of impacted files within software reposi-tories is a key challenge in change impact analysis. Existing deterministic approaches that combine heuristic signals, semantic similarity measures, and graph-based centrality metrics have demonstrated effectiveness in nar-rowing candidate search spaces, yet their recall plateaus. This limitation stems from the treatment of features as linearly independent contributors, ignoring contextual dependencies and relationships between metrics that characterize expert reasoning patterns. To address this limitation, we propose the application of Multi-Head Self-Attention as a post-deterministic scoring refinement mechanism. Our approach learns contextual weighting between features, dynamically adjust-ing importance levels per file based on relational behavior exhibited across candidate file sets. The attention mechanism produces context-aware adjustments that are additively combined with deterministic scores, pre-serving interpretability while enabling reasoning similar to that performed by experts when reviewing change surfaces. We focus on recall rather than precision, as false negatives (missing impacted files) are far more costly than false positives (irrelevant files that can be quickly dismissed during review). Empirical evaluation on 200 test cases demonstrates that the introduc-tion of self-attention improves Top-50 recall from approximately 62-65% to between 78-82% depending on repository complexity and structure, achiev-ing 80% recall at Top-50 files. Expert validation yields improvement from 6.5/10 to 8.6/10 in subjective accuracy alignment. This transformation bridges the reasoning capability gap between deterministic automation and expert judgment, improving recall in repository-aware effort estimation.

cs.SE↗

arXiv2025

Production-Grade Local LLM Inference on Apple Silicon: A Comparative Study of MLX, MLC-LLM, Ollama, llama.cpp, and PyTorch MPS

We present a systematic, empirical evaluation of five local large language model (LLM) runtimes on Apple Silicon: MLX, MLC-LLM, llama.cpp, Ollama, and PyTorch MPS. Experiments were conducted on a Mac Studio equipped with an M2 Ultra processor and 192 GB of unified memory. Using the Qwen-2.5 model family across prompts ranging from a few hundred to 100,000 tokens, we measure time-to-first-token (TTFT), steady-state throughput, latency percentiles, long-context behavior (key-value and prompt caching), quantization support, streaming performance, batching and concurrency behavior, and deployment complexity. Under our settings, MLX achieves the highest sustained generation throughput, while MLC-LLM delivers consistently lower TTFT for moderate prompt sizes and offers stronger out-of-the-box inference features. llama.cpp is highly efficient for lightweight single-stream use, Ollama emphasizes developer ergonomics but lags in throughput and TTFT, and PyTorch MPS remains limited by memory constraints on large models and long contexts. All frameworks execute fully on-device with no telemetry, ensuring strong privacy guarantees. We release scripts, logs, and plots to reproduce all results. Our analysis clarifies the design trade-offs in Apple-centric LLM deployments and provides evidence-based recommendations for interactive and long-context processing. Although Apple Silicon inference frameworks still trail NVIDIA GPU-based systems such as vLLM in absolute performance, they are rapidly maturing into viable, production-grade solutions for private, on-device LLM inference.

cs.AR↗

arXiv2024

From Multiple-Choice to Extractive QA: A Case Study for English and Arabic

The rapid evolution of Natural Language Processing (NLP) has favoured major languages such as English, leaving a significant gap for many others due to limited resources. This is especially evident in the context of data annotation, a task whose importance cannot be underestimated, but which is time-consuming and costly. Thus, any dataset for resource-poor languages is precious, in particular when it is task-specific. Here, we explore the feasibility of repurposing an existing multilingual dataset for a new NLP task: we repurpose a subset of the BELEBELE dataset (Bandarkar et al., 2023), which was designed for multiple-choice question answering (MCQA), to enable the more practical task of extractive QA (EQA) in the style of machine reading comprehension. We present annotation guidelines and a parallel EQA dataset for English and Modern Standard Arabic (MSA). We also present QA evaluation results for several monolingual and cross-lingual QA pairs including English, MSA, and five Arabic dialects. We aim to help others adapt our approach for the remaining 120 BELEBELE language variants, many of which are deemed under-resourced. We also provide a thorough analysis and share insights to deepen understanding of the challenges and opportunities in NLP task reformulation.

cs.CL↗

arXiv2023

Supply chain emission estimation using large language models

Large enterprises face a crucial imperative to achieve the Sustainable Development Goals (SDGs), especially goal 13, which focuses on combating climate change and its impacts. To mitigate the effects of climate change, reducing enterprise Scope 3 (supply chain emissions) is vital, as it accounts for more than 90\% of total emission inventories. However, tracking Scope 3 emissions proves challenging, as data must be collected from thousands of upstream and downstream suppliers.To address the above mentioned challenges, we propose a first-of-a-kind framework that uses domain-adapted NLP foundation models to estimate Scope 3 emissions, by utilizing financial transactions as a proxy for purchased goods and services. We compared the performance of the proposed framework with the state-of-art text classification models such as TF-IDF, word2Vec, and Zero shot learning. Our results show that the domain-adapted foundation model outperforms state-of-the-art text mining techniques and performs as well as a subject matter expert (SME). The proposed framework could accelerate the Scope 3 estimation at Enterprise scale and will help to take appropriate climate actions to achieve SDG 13.

cs.CL↗

arXiv2020

A Framework for Crop Price Forecasting in Emerging Economies by Analyzing the Quality of Time-series Data

Accuracy of crop price forecasting techniques is important because it enables the supply chain planners and government bodies to take appropriate actions by estimating market factors such as demand and supply. In emerging economies such as India, the crop prices at marketplaces are manually entered every day, which can be prone to human-induced errors like the entry of incorrect data or entry of no data for many days. In addition to such human prone errors, the fluctuations in the prices itself make the creation of stable and robust forecasting solution a challenging task. Considering such complexities in crop price forecasting, in this paper, we present techniques to build robust crop price prediction models considering various features such as (i) historical price and market arrival quantity of crops, (ii) historical weather data that influence crop production and transportation, (iii) data quality-related features obtained by performing statistical analysis. We additionally propose a framework for context-based model selection and retraining considering factors such as model stability, data quality metrics, and trend analysis of crop prices. To show the efficacy of the proposed approach, we show experimental results on two crops - Tomato and Maize for 14 marketplaces in India and demonstrate that the proposed approach not only improves accuracy metrics significantly when compared against the standard forecasting techniques but also provides robust models.

stat.AP↗

arXiv2015

Taxonomy grounded aggregation of classifiers with different label sets

We describe the problem of aggregating the label predictions of diverse classifiers using a class taxonomy. Such a taxonomy may not have been available or referenced when the individual classifiers were designed and trained, yet mapping the output labels into the taxonomy is desirable to integrate the effort spent in training the constituent classifiers. A hierarchical taxonomy representing some domain knowledge may be different from, but partially mappable to, the label sets of the individual classifiers. We present a heuristic approach and a principled graphical model to aggregate the label predictions by grounding them into the available taxonomy. Our model aggregates the labels using the taxonomy structure as constraints to find the most likely hierarchically consistent class. We experimentally validate our proposed method on image and text classification tasks.

cs.AI↗