Towards Building Efficient Sentence BERT Models using Layer Pruning

This study examines the effectiveness of layer pruning in creating efficient Sentence BERT (SBERT) models. Our goal is to create smaller sentence embedding models that reduce complexity while maintaining strong embedding similarity. We assess BERT models like Muril and MahaBERT-v2 before and after pruning, comparing them with smaller, scratch-trained models like MahaBERT-Small and MahaBERT-Smaller. Through a two-phase SBERT fine-tuning process involving Natural Language Inference (NLI) and Semantic Textual Similarity (STS), we evaluate the impact of layer reduction on embedding quality. Our findings show that pruned models, despite fewer layers, perform competitively with fully layered versions. Moreover, pruned models consistently outperform similarly sized, scratch-trained models, establishing layer pruning as an effective strategy for creating smaller, efficient embedding models. These results highlight layer pruning as a practical approach for reducing computational demand while preserving high-quality embeddings, making SBERT models more accessible for languages with limited technological resources.

On Importance of Pruning and Distillation for Efficient Low Resource NLP

The rise of large transformer models has revolutionized Natural Language Processing, leading to significant advances in tasks like text classification. However, this progress demands substantial computational resources, escalating training duration, and expenses with larger model sizes. Efforts have been made to downsize and accelerate English models (e.g., Distilbert, MobileBert). Yet, research in this area is scarce for low-resource languages. In this study, we explore the case of the low-resource Indic language Marathi. Leveraging the marathi-topic-all-doc-v2 model as our baseline, we implement optimization techniques to reduce computation time and memory usage. Our focus is on enhancing the efficiency of Marathi transformer models while maintaining top-tier accuracy and reducing computational demands. Using the MahaNews document classification dataset and the marathi-topic-all-doc-v2 model from L3Cube, we apply Block Movement Pruning, Knowledge Distillation, and Mixed Precision methods individually and in combination to boost efficiency. We demonstrate the importance of strategic pruning levels in achieving desired efficiency gains. Furthermore, we analyze the balance between efficiency improvements and environmental impact, highlighting how optimized model architectures can contribute to a more sustainable computational ecosystem. Implementing these techniques on a single GPU system, we determine that the optimal configuration is 25\% pruning + knowledge distillation. This approach yielded a 2.56x speedup in computation time while maintaining baseline accuracy levels.

L3Cube-IndicQuest: A Benchmark Questing Answering Dataset for Evaluating Knowledge of LLMs in Indic Context

Large Language Models (LLMs) have made significant progress in incorporating Indic languages within multilingual models. However, it is crucial to quantitatively assess whether these languages perform comparably to globally dominant ones, such as English. Currently, there is a lack of benchmark datasets specifically designed to evaluate the regional knowledge of LLMs in various Indic languages. In this paper, we present the L3Cube-IndicQuest, a gold-standard question-answering benchmark dataset designed to evaluate how well multilingual LLMs capture regional knowledge across various Indic languages. The dataset contains 200 question-answer pairs, each for English and 19 Indic languages, covering five domains specific to the Indic region. We aim for this dataset to serve as a benchmark, providing ground truth for evaluating the performance of LLMs in understanding and representing knowledge relevant to the Indian context. The IndicQuest can be used for both reference-based evaluation and LLM-as-a-judge evaluation. The dataset is shared publicly at https://github.com/l3cube-pune/indic-nlp .

Chain-of-Translation Prompting (CoTR): A Novel Prompting Technique for Low Resource Languages

This paper introduces Chain of Translation Prompting (CoTR), a novel strategy designed to enhance the performance of language models in low-resource languages. CoTR restructures prompts to first translate the input context from a low-resource language into a higher-resource language, such as English. The specified task like generation, classification, or any other NLP function is then performed on the translated text, with the option to translate the output back to the original language if needed. All these steps are specified in a single prompt. We demonstrate the effectiveness of this method through a case study on the low-resource Indic language Marathi. The CoTR strategy is applied to various tasks, including sentiment analysis, hate speech classification, subject classification and text generation, and its efficacy is showcased by comparing it with regular prompting methods. Our results underscore the potential of translation-based prompting strategies to significantly improve multilingual LLM performance in low-resource languages, offering valuable insights for future research and applications. We specifically see the highest accuracy improvements with the hate speech detection task. The technique also has the potential to enhance the quality of synthetic data generation for underrepresented languages using LLMs.

A Data Selection Approach for Enhancing Low Resource Machine Translation Using Cross-Lingual Sentence Representations

Machine translation in low-resource language pairs faces significant challenges due to the scarcity of parallel corpora and linguistic resources. This study focuses on the case of English-Marathi language pairs, where existing datasets are notably noisy, impeding the performance of machine translation models. To mitigate the impact of data quality issues, we propose a data filtering approach based on cross-lingual sentence representations. Our methodology leverages a multilingual SBERT model to filter out problematic translations in the training data. Specifically, we employ an IndicSBERT similarity model to assess the semantic equivalence between original and translated sentences, allowing us to retain linguistically correct translations while discarding instances with substantial deviations. The results demonstrate a significant improvement in translation quality over the baseline post-filtering with IndicSBERT. This illustrates how cross-lingual sentence representations can reduce errors in machine translation scenarios with limited resources. By integrating multilingual sentence BERT models into the translation pipeline, this research contributes to advancing machine translation techniques in low-resource environments. The proposed method not only addresses the challenges in English-Marathi language pairs but also provides a valuable framework for enhancing translation quality in other low-resource language translation tasks.

LLM Pruning and Distillation in Practice: The Minitron Approach

We present a comprehensive report on compressing the Llama 3.1 8B and Mistral NeMo 12B models to 4B and 8B parameters, respectively, using pruning and distillation. We explore two distinct pruning strategies: (1) depth pruning and (2) joint hidden/attention/MLP (width) pruning, and evaluate the results on common benchmarks from the LM Evaluation Harness. The models are then aligned with NeMo Aligner and tested in instruct-tuned versions. This approach produces a compelling 4B model from Llama 3.1 8B and a state-of-the-art Mistral-NeMo-Minitron-8B (MN-Minitron-8B for brevity) model from Mistral NeMo 12B. We found that with no access to the original data, it is beneficial to slightly fine-tune teacher models on the distillation dataset. We open-source our base model weights on Hugging Face with a permissive license.

Leveraging Parameter Efficient Training Methods for Low Resource Text Classification: A Case Study in Marathi

With the surge in digital content in low-resource languages, there is an escalating demand for advanced Natural Language Processing (NLP) techniques tailored to these languages. BERT (Bidirectional Encoder Representations from Transformers), serving as the foundational framework for numerous NLP architectures and language models, is increasingly employed for the development of low-resource NLP models. Parameter Efficient Fine-Tuning (PEFT) is a method for fine-tuning Large Language Models (LLMs) and reducing the training parameters to some extent to decrease the computational costs needed for training the model and achieve results comparable to a fully fine-tuned model. In this work, we present a study of PEFT methods for the Indic low-resource language Marathi. We conduct a comprehensive analysis of PEFT methods applied to various monolingual and multilingual Marathi BERT models. These approaches are evaluated on prominent text classification datasets like MahaSent, MahaHate, and MahaNews. The incorporation of PEFT techniques is demonstrated to significantly expedite the training speed of the models, addressing a critical aspect of model development and deployment. In this study, we explore Low-Rank Adaptation of Large Language Models (LoRA) and adapter methods for low-resource text classification. We show that these methods are competitive with full fine-tuning and can be used without loss in accuracy. This study contributes valuable insights into the effectiveness of Marathi BERT models, offering a foundation for the continued advancement of NLP capabilities in Marathi and similar Indic languages.

Compact Language Models via Pruning and Knowledge Distillation

Large language models (LLMs) targeting different deployment scales and sizes are currently produced by training each variant from scratch; this is extremely compute-intensive. In this paper, we investigate if pruning an existing LLM and then re-training it with a fraction (<3%) of the original training data can be a suitable alternative to repeated, full retraining. To this end, we develop a set of practical and effective compression best practices for LLMs that combine depth, width, attention and MLP pruning with knowledge distillation-based retraining; we arrive at these best practices through a detailed empirical exploration of pruning strategies for each axis, methods to combine axes, distillation strategies, and search techniques for arriving at optimal compressed architectures. We use this guide to compress the Nemotron-4 family of LLMs by a factor of 2-4x, and compare their performance to similarly-sized models on a variety of language modeling tasks. Deriving 8B and 4B models from an already pretrained 15B model using our approach requires up to 40x fewer training tokens per model compared to training from scratch; this results in compute cost savings of 1.8x for training the full model family (15B, 8B, and 4B). Minitron models exhibit up to a 16% improvement in MMLU scores compared to training from scratch, perform comparably to other community models such as Mistral 7B, Gemma 7B and Llama-3 8B, and outperform state-of-the-art compression techniques from the literature. We have open-sourced Minitron model weights on Huggingface, with corresponding supplementary material including example code available on GitHub.

Universal Cross-Lingual Text Classification

Text classification, an integral task in natural language processing, involves the automatic categorization of text into predefined classes. Creating supervised labeled datasets for low-resource languages poses a considerable challenge. Unlocking the language potential of low-resource languages requires robust datasets with supervised labels. However, such datasets are scarce, and the label space is often limited. In our pursuit to address this gap, we aim to optimize existing labels/datasets in different languages. This research proposes a novel perspective on Universal Cross-Lingual Text Classification, leveraging a unified model across languages. Our approach involves blending supervised data from different languages during training to create a universal model. The supervised data for a target classification task might come from different languages covering different labels. The primary goal is to enhance label and language coverage, aiming for a label set that represents a union of labels from various languages. We propose the usage of a strong multilingual SBERT as our base model, making our novel training strategy feasible. This strategy contributes to the adaptability and effectiveness of the model in cross-lingual language transfer scenarios, where it can categorize text in languages not encountered during training. Thus, the paper delves into the intricacies of cross-lingual text classification, with a particular focus on its application for low-resource languages, exploring methodologies and implications for the development of a robust and adaptable universal cross-lingual model.

Curating Stopwords in Marathi: A TF-IDF Approach for Improved Text Analysis and Information Retrieval

Stopwords are commonly used words in a language that are often considered to be of little value in determining the meaning or significance of a document. These words occur frequently in most texts and don't provide much useful information for tasks like sentiment analysis and text classification. English, which is a high-resource language, takes advantage of the availability of stopwords, whereas low-resource Indian languages like Marathi are very limited, standardized, and can be used in available packages, but the number of available words in those packages is low. Our work targets the curation of stopwords in the Marathi language using the MahaCorpus, with 24.8 million sentences. We make use of the TF-IDF approach coupled with human evaluation to curate a strong stopword list of 400 words. We apply the stop word removal to the text classification task and show its efficacy. The work also presents a simple recipe for stopword curation in a low-resource language. The stopwords are integrated into the mahaNLP library and publicly available on https://github.com/l3cube-pune/MarathiNLP .