Ashutosh Hathidara

Vaishali Senthil, Ashutosh Hathidara, Sebastian Schreiber

Tool retrieval over large API catalogs is a core bottleneck for LLM agents: user queries arrive in colloquial, often underspecified language, while the catalog uses technical API vocabulary that no fixed encoder can bridge on its own. The two dominant training approaches, contrastive encoder fine-tuning and HyDE-style query expansion with a frozen LLM, address this problem from opposite ends and fail in complementary directions: the fine-tuned encoder excels when the query's surface form already matches the catalog but collapses when it does not, while zero-shot HyDE is more robust to underspecified queries yet generates catalog-unaware hypothetical descriptions that degrade retrieval when queries are well-formed. We introduce CoHyDE, an iterative procedure that trains the dense encoder and the LLM rewriter as a single co-evolving system: the encoder is retrained with InfoNCE on catalog-style hypothetical descriptions produced by the rewriter, and the rewriter is preference-aligned via DPO against the encoder's retrieval scores, with both sides warm-started on the tool catalog before the loop begins. On a ~10k tool subset of the ToolBench catalog, three rounds of CoHyDE improve over the strongest single-component baseline by +2.5 pp NDCG@5 on standard queries and +6.3 pp on held-out vague queries, with gains as large as +8 pp on the hardest vague tier. Ablations confirm that co-training is the key ingredient: using either component in isolation fails to match CoHyDE on both well-formed and vague queries, with losses of up to -8 pp on vague queries.

Ashutosh Hathidara, Julien Yu, Vaishali Senthil et al.

Large language models (LLMs) are increasingly used as human simulators, both for evaluating conversational systems and for generating fine-tuning data. However, naive "act-as-a-user" prompting often yields verbose, unrealistic utterances, underscoring the need for principled evaluation of so-called user proxy agents. We present MIRRORBENCH, a reproducible, extensible benchmarking framework that evaluates user proxies solely on their ability to produce human-like user utterances across diverse conversational tasks, explicitly decoupled from downstream task success. MIRRORBENCH features a modular execution engine with typed interfaces, metadata-driven registries, multi-backend support, caching, and robust observability. The system supports pluggable user proxies, datasets, tasks, and metrics, enabling researchers to evaluate arbitrary simulators under a uniform, variance-aware harness. We include three lexical-diversity metrics (MATTR, YULE'S K, and HD-D) and three LLM-judge-based metrics (GTEval, Pairwise Indistinguishability, and Rubric-and-Reason). Across four open datasets, MIRRORBENCH yields variance-aware results and reveals systematic gaps between user proxies and real human users. The framework is open source and includes a simple command-line interface for running experiments, managing configurations and caching, and generating reports. The framework can be accessed at https://github.com/SAP/mirrorbench.

Ashutosh Hathidara, Julien Yu, Sebastian Schreiber

Large language models (LLMs) are increasingly tasked with invoking enterprise APIs, yet they routinely falter when near-duplicate tools vie for the same user intent or when required arguments are left underspecified. We introduce DiaFORGE (Dialogue Framework for Organic Response Generation & Evaluation), a disambiguation-centric, three-stage pipeline that (i) synthesizes persona-driven, multi-turn dialogues in which the assistant must distinguish among highly similar tools, (ii) performs supervised fine-tuning of open-source models with reasoning traces across 3B - 70B parameters, and (iii) evaluates real-world readiness via a dynamic suite that redeploys each model in a live agentic loop and reports end-to-end goal completion alongside conventional static metrics. On our dynamic benchmark DiaBENCH, models trained with DiaFORGE raise tool-invocation success by 27 pp over GPT-4o and by 49 pp over Claude-3.5-Sonnet, both under optimized prompting. To spur further research, we release an open corpus of 5000 production-grade enterprise API specifications paired with rigorously validated, disambiguation-focused dialogues, offering a practical blueprint for building reliable, enterprise-ready tool-calling agents.

Ashutosh Hathidara, Lalit Pandey

A Perceptron is a fundamental building block of a neural network. The flexibility and scalability of perceptron make it ubiquitous in building intelligent systems. Studies have shown the efficacy of a single neuron in making intelligent decisions. Here, we examined and compared two perceptrons with distinct mechanisms, and developed a quantum version of one of those perceptrons. As a part of this modeling, we implemented the quantum circuit for an artificial perception, generated a dataset, and simulated the training. Through these experiments, we show that there is an exponential growth advantage and test different qubit versions. Our findings show that this quantum model of an individual perceptron can be used as a pattern classifier. For the second type of model, we provide an understanding to design and simulate a spike-dependent quantum perceptron. Our code is available at https://github.com/ashutosh1919/quantum-perceptron

Ashutosh Hathidara, Lalit Pandey

Deep Neural Networks have achieved great success in some of the complex tasks that humans can do with ease. These include image recognition/classification, natural language processing, game playing etc. However, modern Neural Networks fail or perform poorly when trained on tasks that can be solved easily using backtracking and traditional algorithms. Therefore, we use the architecture of the Neuro Logic Machine (NLM) and extend its functionality to solve a 9X9 game of Sudoku. To expand the application of NLMs, we generate a random grid of cells from a dataset of solved games and assign up to 10 new empty cells. The goal of the game is then to find a target value ranging from 1 to 9 and fill in the remaining empty cells while maintaining a valid configuration. In our study, we showcase an NLM which is capable of obtaining 100% accuracy for solving a Sudoku with empty cells ranging from 3 to 10. The purpose of this study is to demonstrate that NLMs can also be used for solving complex problems and games like Sudoku. We also analyze the behaviour of NLMs with a backtracking algorithm by comparing the convergence time using a graph plot on the same problem. With this study we show that Neural Logic Machines can be trained on the tasks that traditional Deep Learning architectures fail using Reinforcement Learning. We also aim to propose the importance of symbolic learning in explaining the systematicity in the hybrid model of NLMs.

Ashutosh Hathidara, Gaurav Atavale, Suyash Chaudhary

Bitcoin has increased investment interests in people during the last decade. We have seen an increase in the number of posts on social media platforms about cryptocurrency, especially Bitcoin. This project focuses on analyzing user tweet data in combination with Bitcoin price data to see the relevance between price fluctuations and the conversation between millions of people on Twitter. This study also exploits this relationship between user tweets and bitcoin prices to predict the future bitcoin price. We are utilizing novel techniques and methods to analyze the data and make price predictions.