In the context of pretraining of Large Language Models (LLMs), synthetic data has emerged as an alternative for generating high-quality pretraining data at scale. This is particularly beneficial in low resource language settings where the benefits of the recent LLMs have been unevenly distributed across languages. In this work, we present a systematic study on the generation and evaluation of synthetic multilingual pretraining data for Indic languages, where we construct a large scale synthetic dataset BhashaKritika, comprising 540B tokens using 5 different techniques for 10 languages. We explore the impact of grounding generation in documents, personas and topics. We analyze how language choice, both in the prompt instructions and document grounding affects data quality and we compare translations of English content with native generation in Indic languages. In order to support scalable and language-sensitive evaluation, we introduce a modular quality evaluation pipeline that integrates script and language detection, metadata consistency checks, n-gram repetition analysis, and perplexity-based filtering using KenLM models. Our framework enables robust quality control across diverse scripts and linguistic contexts. Empirical results through model runs reveal key trade-offs in generation strategies and highlight best practices for constructing effective multilingual corpora. This work contributes practical insights for advanced pretraining recipes in low-resource and script-diverse settings, particularly in the Indian context.

Tokenizers play a crucial role in determining the performance, training efficiency, and the inference cost of Large Language Models (LLMs).Designing effective tokenizers for multilingual LLMs is particularly challenging due to diverse scripts and rich morphological variation.While subword methods like Byte Pair Encoding (BPE) are widely adopted, their effectiveness in multilingual settings remains underexplored.We present MUTANT, a recipe for building multilingual tokenizers, with careful vocabulary and training data design, language-aware pre-tokenization, and subword and multiword aware training.We also introduce MUTANT-Indic, a tokenizer for Indiaspecific multilingual LLMs, that produces linguistically coherent tokens and achieves stateof-the-art performance.Evaluated across English, 22 Indian languages and code data, our tokenizer improves the average fertility score by 39.5% over LLaMA4 and by 18% over Sutra (the current best).This translates to 44% improvement in inference throughput over LLaMA4 while maintaining comparable performance on English and Indic benchmarks.We present detailed ablations across tokenizer training data size, vocabulary size, merging techniques, and pre-tokenization strategies, demonstrating the robustness of our design choices.

Recent multimodal foundation models are primarily trained on English or high resource European language data, which limits their applicability to other medium and low-resource languages, such as the Indian languages. To address this limitation, we introduce Chitrarth (Chitra: Image; Artha: Meaning), an inclusive Vision-Language Model (VLM), specifically targeting the rich linguistic diversity and visual reasoning across 10 prominent Indian languages. Our model effectively integrates a state-of-the-art (SOTA) multilingual Large Language Model (LLM) with a vision module, primarily trained on multilingual image-text data. Furthermore, we also introduce BharatBench, a comprehensive framework for evaluating VLMs across various low resource languages, ultimately contributing to more diverse and effective AI systems. Our model presents SOTA results for benchmarks across Indian languages while retaining its efficiency in English. Through our research, we aim to set new benchmarks in multilingual-multimodal capabilities, offering substantial improvements over existing models and establishing a foundation for facilitating future advancements in this arena.

The quality of automatic speech recognition (ASR) is critical to Dialogue Systems as ASR errors propagate to and directly impact downstream tasks such as language understanding (LU). In this paper, we propose multi-task neural approaches to perform contextual language correction on ASR outputs jointly with LU to improve the performance of both tasks simultaneously. To measure the effectiveness of this approach we used a public benchmark, the 2nd Dialogue State Tracking (DSTC2) corpus. As a baseline approach, we trained task specific Statistical Language Models (SLM) and fine-tuned state-of-the-art Generative Pre-training (GPT) Language Model to re-rank the n-best ASR hypotheses, followed by a model to identify the dialog act and slots. i) We further trained ranker models using GPT and Hierarchical CNN-RNN models with discriminatory losses to detect the best output given n-best hypotheses. We extended these ranker models to first select the best ASR output and then identify the dialogue act and slots in an end to end fashion. ii) We also proposed a novel joint ASR error correction and LU model, a word confusion pointer network (WCN-Ptr) with multihead self attention on top, which consumes the word confusions populated from the n-best. We show that the error rates of off the shelf ASR and following LU systems can be reduced significantly by 14% relative with joint models trained using small amounts of in-domain data.

This work presents an exploration and imitation-learning-based agent capable of state-of-the-art performance in playing text-based computer games. These games are of interest as they can be seen as a testbed for language understanding, problem-solving, and language generation by artificial agents. Moreover, they provide a learning setting in which these skills can be acquired through interactions with an environment rather than using fixed corpora. One aspect that makes these games particularly challenging for learning agents is the combinatorially large action space. Existing methods for solving text-based games are limited to games that are either very simple or have an action space restricted to a predetermined set of admissible actions. In this work, we propose to use the exploration approach of Go-Explore for solving text-based games. More specifically, in an initial exploration phase, we first extract trajectories with high rewards, after which we train a policy to solve the game by imitating these trajectories. Our experiments show that this approach outperforms existing solutions in solving text-based games, and it is more sample efficient in terms of the number of interactions with the environment. Moreover, we show that the learned policy can generalize better than existing solutions to unseen games without using any restriction on the action space.

Alessandra Cervone, Chandra Khatri, Rahul Goel, Behnam Hedayatnia, Anu Venkatesh, Dilek Hakkani-Tur, Raefer Gabriel. Proceedings of the 12th International Conference on Natural Language Generation. 2019.

Sanghyun Yi, Rahul Goel, Chandra Khatri, Alessandra Cervone, Tagyoung Chung, Behnam Hedayatnia, Anu Venkatesh, Raefer Gabriel, Dilek Hakkani-Tur. Proceedings of the 12th International Conference on Natural Language Generation. 2019.

The first workshop on Conversational Interaction Systems is held in Paris, France on July 25th, 2019, co-located with the ACM Special Interest Group on Information Retrieval (SIGIR). The goal of the workshop is to bring together researchers from academia and industry to discuss the challenges and future of conversational agents and interactive systems. The workshop has an exciting program that spans a number of subareas including: multi-modal conversational interfaces, dialogue accessibility, and scaling such systems. The program includes eight invited talks, a lively panel discussion on emerging topics, and presentation of original research papers.

Statistical language models (LM) play a key role in Automatic Speech Recognition (ASR) systems used by conversational agents. These ASR systems should provide a high accuracy under a variety of speaking styles, domains, vocabulary and argots. In this paper, we present a DNN-based method to adapt the LM to each user-agent interaction based on generalized contextual information, by predicting an optimal, context-dependent set of LM interpolation weights. We show that this framework for contextual adaptation provides accuracy improvements under different possible mixture LM partitions that are relevant for both (1) Goal-oriented conversational agents where it's natural to partition the data by the requested application and for (2) Non-goal oriented conversational agents where the data can be partitioned using topic labels that come from predictions of a topic classifier. We obtain a relative WER improvement of 3% with a 1-pass decoding strategy and 6% in a 2-pass decoding framework, over an unadapted model. We also show up to a 15% relative improvement in recognizing named entities which is of significant value for conversational ASR systems.

Accurate prediction of conversation topics can be a valuable signal for creating coherent and engaging dialog systems. In this work, we focus on context-aware topic classification methods for identifying topics in free-form human-chatbot dialogs. We extend previous work on neural topic classification and unsupervised topic keyword detection by incorporating conversational context and dialog act features. On annotated data, we show that incorporating context and dialog acts leads to relative gains in topic classification accuracy by 35% and on unsupervised keyword detection recall by 11% for conversational interactions where topics frequently span multiple utterances. We show that topical metrics such as topical depth is highly correlated with dialog evaluation metrics such as coherence and engagement implying that conversational topic models can predict user satisfaction. Our work for detecting conversation topics and keywords can be used to guide chatbots towards coherent dialog.

Content is one of the most essential parts of products on e-commerce websites such as eBay. It not only drives user-engagement but also traffic from various search engine websites based on the relevance. Generating the content for the products, however comes with a wide set of challenges, due to the complexity of commerce at scale, and requires new applications in text processing and information extraction to address some core issues. Some of the factors which need to be addressed are: scalability (millions of products), dynamism (products change with time), removal of item-specific or seller specific information (maintain generality), size of the content etc. Generally, curators are hired for writing the product descriptions manually, which is not cost-effective and is not scalable. In the current work, an algorithmic framework based on Natural Language Processing and Deep Learning is proposed and used to generate the content for ecommerce products. Seller descriptions for multiple items aggregated at a product level are used for content generation. Furthermore, a combination of behavioral and text signals such as search queries are also used to understand the user intent. Two different approaches are proposed in this work: Extraction (sentence retrieval) and Abstraction (sentence generation). The results of both the methods are analyzed and it is depicted that algorithmic content generation is scalable, fast and has potential to cut down the manualcuration cost dramatically.