Enhancing Q&A Text Retrieval with Ranking Models: Benchmarking, fine-tuning and deploying Rerankers for RAG

Ranking models play a crucial role in enhancing overall accuracy of text retrieval systems. These multi-stage systems typically utilize either dense embedding models or sparse lexical indices to retrieve relevant passages based on a given query, followed by ranking models that refine the ordering of the candidate passages by its relevance to the query. This paper benchmarks various publicly available ranking models and examines their impact on ranking accuracy. We focus on text retrieval for question-answering tasks, a common use case for Retrieval-Augmented Generation systems. Our evaluation benchmarks include models some of which are commercially viable for industrial applications. We introduce a state-of-the-art ranking model, NV-RerankQA-Mistral-4B-v3, which achieves a significant accuracy increase of ~14% compared to pipelines with other rerankers. We also provide an ablation study comparing the fine-tuning of ranking models with different sizes, losses and self-attention mechanisms. Finally, we discuss challenges of text retrieval pipelines with ranking models in real-world industry applications, in particular the trade-offs among model size, ranking accuracy and system requirements like indexing and serving latency / throughput.

Winning Amazon KDD Cup'24

This paper describes the winning solution of all 5 tasks for the Amazon KDD Cup 2024 Multi Task Online Shopping Challenge for LLMs. The challenge was to build a useful assistant, answering questions in the domain of online shopping. The competition contained 57 diverse tasks, covering 5 different task types (e.g. multiple choice) and across 4 different tracks (e.g. multi-lingual). Our solution is a single model per track. We fine-tune Qwen2-72B-Instruct on our own training dataset. As the competition released only 96 example questions, we developed our own training dataset by processing multiple public datasets or using Large Language Models for data augmentation and synthetic data generation. We apply wise-ft to account for distribution shifts and ensemble multiple LoRA adapters in one model. We employed Logits Processors to constrain the model output on relevant tokens for the tasks. AWQ 4-bit Quantization and vLLM are used during inference to predict the test dataset in the time constraints of 20 to 140 minutes depending on the track. Our solution achieved the first place in each individual track and is the first place overall of Amazons KDD Cup 2024.

NV-Retriever: Improving text embedding models with effective hard-negative mining

Text embedding models have been popular for information retrieval applications such as semantic search and Question-Answering systems based on Retrieval-Augmented Generation (RAG). Those models are typically Transformer models that are fine-tuned with contrastive learning objectives. Many papers introduced new embedding model architectures and training approaches, however, one of the key ingredients, the process of mining negative passages, remains poorly explored or described. One of the challenging aspects of fine-tuning embedding models is the selection of high quality hard-negative passages for contrastive learning. In this paper we propose a family of positive-aware mining methods that leverage the positive relevance score for more effective false negatives removal. We also provide a comprehensive ablation study on hard-negative mining methods over their configurations, exploring different teacher and base models. We demonstrate the efficacy of our proposed methods by introducing the NV-Retriever-v1 model, which scores 60.9 on MTEB Retrieval (BEIR) benchmark and 0.65 points higher than previous methods. The model placed 1st when it was published to MTEB Retrieval on July 07, 2024.

Robust Temporal Ensembling for Learning with Noisy Labels

Successful training of deep neural networks with noisy labels is an essential capability as most real-world datasets contain some amount of mislabeled data. Left unmitigated, label noise can sharply degrade typical supervised learning approaches. In this paper, we present robust temporal ensembling (RTE), which combines robust loss with semi-supervised regularization methods to achieve noise-robust learning. We demonstrate that RTE achieves state-of-the-art performance across the CIFAR-10, CIFAR-100, ImageNet, WebVision, and Food-101N datasets, while forgoing the recent trend of label filtering and/or fixing. Finally, we show that RTE also retains competitive corruption robustness to unforeseen input noise using CIFAR-10-C, obtaining a mean corruption error (mCE) of 13.50% even in the presence of an 80% noise ratio, versus 26.9% mCE with standard methods on clean data.

A Sensorized Multicurved Robot Finger with Data-driven Touch Sensing via Overlapping Light Signals

Despite significant advances in touch and force transduction, tactile sensing is still far from ubiquitous in robotic manipulation. Existing methods for building touch sensors have proven difficult to integrate into robot fingers due to multiple challenges, including difficulty in covering multicurved surfaces, high wire count, or packaging constrains preventing their use in dexterous hands. In this paper, we present a multicurved robotic finger with accurate touch localization and normal force detection over complex, three-dimensional surfaces. The key to our approach is the novel use of overlapping signals from light emitters and receivers embedded in a transparent waveguide layer that covers the functional areas of the finger. By measuring light transport between every emitter and receiver, we show that we can obtain a very rich signal set that changes in response to deformation of the finger due to touch. We then show that purely data-driven deep learning methods are able to extract useful information from such data, such as contact location and applied normal force, without the need for analytical models. The final result is a fully integrated, sensorized robot finger, with a low wire count and using easily accessible manufacturing methods, designed for easy integration into dexterous manipulators.

Accurate Trajectory Prediction for Autonomous Vehicles

Predicting vehicle trajectories, angle and speed is important for safe and comfortable driving. We demonstrate the best predicted angle, speed, and best performance overall winning the top three places of the ICCV 2019 Learning to Drive challenge. Our key contributions are (i) a general neural network system architecture which embeds and fuses together multiple inputs by encoding, and decodes multiple outputs using neural networks, (ii) using pre-trained neural networks for augmenting the given input data with segmentation maps and semantic information, and (iii) leveraging the form and distribution of the expected output in the model.