Publications

2026

1. 

   Application of LLM Agents for Long-Form Question Answering in Systems Biology

   Charvi Jain, Sahar Vahdati, Pauline Porschitz, and 4 more authors

   In Machine Learning and Knowledge Discovery in Databases. Applied Data Science Track, Sep 2026

   Abs PDF Poster Slides

   AI-based research assistants for domains with cross disciplinary and multi-resource demands, such as systems biology, require deep reasoning and integration across diverse knowledge sources. In this work, we study the application of large language model (LLM) agents for long-form scientific question answering that requires multi-source reasoning across heterogeneous biological knowledge resources. To achieve our aim, we first introduce a high-quality, human-curated dataset, ThinkSysBio, that offers a structured testbed for long-form question answering (LFQA) with multi-source, tool-augmented scientific reasoning in systems biology. Second, we devise an adaptation of the ReAct framework, namely ReAct-LF, for inference tokens efficient LFQA. In addition, we evaluate both ReAct and ReAct-LF using four state-of-the-art open-source and closed-source LLMs. Third, we perform a systematic evaluation of agentic workflows by analyzing both the intermediate reasoning traces and the generated scholarly reports. Our results reveal a practical trade-off between report quality and computational efficiency in agent-based scientific workflows, demonstrating that relevance-aware evidence filtering can substantially reduce inference token usage by an average of 61.3% across datasets while maintaining comparable report quality.

2024

1. 

   Evaluating Large Language Model Literature Reviews In Interdisciplinary Science: A Systems Biology Perspective

   Charvi Jain, Sahar Vahdati, Nandu Gopan, and 2 more authors

   Nov 2024

   Abs PDF Poster

   We evaluate the effectiveness of current large language model (LLM) literature review systems in interdisciplinary domains. While LLMs can support and accelerate reviewing the scientific literature, it is unclear how they cope with interdisciplinary science, where sources from multiple fields must be integrated according to relevance defined by context. We study this from the perspective of systems biology, a field that combines biology, mathematics, physics, and computer science. Using a set of expert-defined research questions, we assess the ability of LLMs to meaningfully integrate cross-domain knowledge and correctly reflect relevance. Specifically, we evaluate the quality of generated reports and the relevance of retrieved references from five different review models. We find that LLMs are a valuable augmentative tool for literature reviews, but trade off report quality for completeness in interdisciplinary domains. We address these limitations by proposing a novel method, termed AURORA, which is particularly designed for interdisciplinary applications. On the interdisciplinary systems biology benchmark, AURORA offers good coverage with high-quality reports.

2. 

   Teuken-7B-Base & Teuken-7B-Instruct: Towards European LLMs

   Mehdi Ali, Michael Fromm, Klaudia Thellmann, and 36 more authors

   Oct 2024

   Abs PDF

   We present two multilingual LLMs designed to embrace Europe’s linguistic diversity by supporting all 24 official languages of the European Union. Trained on a dataset comprising around 60% non-English data and utilizing a custom multilingual tokenizer, our models address the limitations of existing LLMs that predominantly focus on English or a few high-resource languages. We detail the models’ development principles, i.e., data composition, tokenizer optimization, and training methodologies. The models demonstrate competitive performance across multilingual benchmarks, as evidenced by their performance on European versions of ARC, HellaSwag, MMLU, and TruthfulQA.

3. 

   Tokenizer Choice For LLM Training: Negligible or Crucial?

   Mehdi Ali, Michael Fromm, Klaudia Thellmann, and 18 more authors

   In Findings of the Association for Computational Linguistics: NAACL 2024, Jun 2024

   Abs PDF

   The recent success of large language models (LLMs) has been predominantly driven by curating the training dataset composition, scaling of model architectures and dataset sizes and advancements in pretraining objectives, leaving tokenizer influence as a blind spot.Shedding light on this underexplored area, we conduct a comprehensive study on the influence of tokenizer choice on LLM downstream performance by training 24 mono- and multilingual LLMs at a 2.6B parameter scale, ablating different tokenizer algorithms and parameterizations. Our studies highlight that the tokenizer choice can significantly impact the model’s downstream performance and training costs. In particular, we find that the common tokenizer evaluation metrics fertility and parity are not always predictive of model downstream performance, rendering these metrics a questionable proxy for the model’s downstream performance. Furthermore, we show that multilingual tokenizers trained on the five most frequent European languages require vocabulary size increases of factor three in comparison to English. While English-centric tokenizers have been applied to the training of multi-lingual LLMs in the past, we find that this approach results in a severe downstream performance degradation and additional training costs of up to 68%, due to an inefficient tokenization vocabulary.

2019

1. Characterization of facial expression using deep neural networks

   Neha Sharma, and Charvi Jain

   In 2019 5th International Conference on Advanced Computing & Communication Systems (ICACCS), Jun 2019
2. Image Captioning Using CNN Long Short-Term Memory Network

   Charvi Jain, and Neha Sharma

   In Proceedings of the International Conference on Advances in Electronics, Electrical & Computational Intelligence (ICAEEC), Jun 2019

2018

1. Abnormal behaviour detection at traffic junctions using Lucas Kanade and Harris Corner detector

   Charvi Jain, and Diwakar Gautam

   In 2018 4th International Conference on Recent Advances in Information Technology (RAIT), Jun 2018
2. Strengthening of Fingerprint Technology

   Charvi Jain

   In 2018 International Conference on Computing, Power and Communication Technologies (GUCON), Jun 2018
3. Emotion detection and characterization using facial features

   Charvi Jain, Kshitij Sawant, Mohammed Rehman, and 1 more author

   In 2018 3rd International Conference and Workshops on Recent Advances and innovations in Engineering (ICRAIE), Jun 2018