XFBoost: Improving Text Generation with Controllable Decoders

TL;DR

XFBoost is a controllable text generation framework that enhances product descriptions by integrating visual attributes as constraints and fine-tuning with policy gradients, resulting in more accurate and relevant outputs.

Contribution

The paper introduces XFBoost, a novel framework combining visual constraints and policy gradient fine-tuning to improve the factual accuracy and relevance of generated product descriptions.

Findings

Significantly improves description relevance and accuracy.

Reduces factual inaccuracies in generated texts.

Effective in online learning with human feedback.

Abstract

Multimodal conditionality in transformer-based natural language models has demonstrated state-of-the-art performance in the task of product description generation. Recent approaches condition a language model on one or more images and other textual metadata to achieve near-human performance for describing products from e-commerce stores. However, generated descriptions may exhibit degrees of inaccuracy or even contradictory claims relative to the inputs of a given product. In this paper, we propose a controllable language generation framework called Extract-Finetune-Boost (XFBoost), which addresses the problem of inaccurate low-quality inference. By using visual semantic attributes as constraints at the decoding stage of the generation process and finetuning the language model with policy gradient techniques, the XFBoost framework is found to produce significantly more descriptive text with higher image relevancy, outperforming baselines and lowering the frequency of factually inaccurate descriptions. We further demonstrate the application of XFBoost to online learning wherein human-in-the-loop critics improve language models with active feedback.