Joseph Liu

Mahesh Kumar Nandwana, Youngwan Lim, Joseph Liu et al.

Large Language Models (LLMs) are typically aligned for safety during the post-training phase; however, they may still generate inappropriate outputs that could potentially pose risks to users. This challenge underscores the need for robust safeguards that operate across both model inputs and outputs. In this work, we introduce Roblox Guard 1.0, a state-of-the-art instruction fine-tuned LLM designed to enhance the safety of LLM systems through comprehensive input-output moderation, using a pipeline of LLMs to enhance moderation capability. Built on the Llama-3.1-8B-Instruct backbone, our model is instruction fine-tuned to generalize across previously unseen safety taxonomies and demonstrates strong performance on out-of-domain safety benchmarks. The instruction fine-tuning process uses a mix of synthetic and open-source safety datasets, augmented with chain-of-thought (CoT) rationales and input inversion to enhance contextual understanding and decision making. To support systematic evaluation, we also release RobloxGuard-Eval, a new benchmark featuring an extensible safety taxonomy to assess the effectiveness of LLM guardrails and moderation frameworks.

Joseph Liu, Joshua Geddes, Ziyu Guo et al.

Diffusion Transformers (DiT) have emerged as powerful generative models for various tasks, including image, video, and speech synthesis. However, their inference process remains computationally expensive due to the repeated evaluation of resource-intensive attention and feed-forward modules. To address this, we introduce SmoothCache, a model-agnostic inference acceleration technique for DiT architectures. SmoothCache leverages the observed high similarity between layer outputs across adjacent diffusion timesteps. By analyzing layer-wise representation errors from a small calibration set, SmoothCache adaptively caches and reuses key features during inference. Our experiments demonstrate that SmoothCache achieves 8% to 71% speed up while maintaining or even improving generation quality across diverse modalities. We showcase its effectiveness on DiT-XL for image generation, Open-Sora for text-to-video, and Stable Audio Open for text-to-audio, highlighting its potential to enable real-time applications and broaden the accessibility of powerful DiT models.

Joseph Liu, Mahesh Kumar Nandwana, Janne Pylkkönen et al.

Toxicity classification for voice heavily relies on the semantic content of speech. We propose a novel framework that utilizes cross-modal learning to integrate the semantic embedding of text into a multilabel speech toxicity classifier during training. This enables us to incorporate textual information during training while still requiring only audio during inference. We evaluate this classifier on large-scale datasets with real-world characteristics to validate the effectiveness of this framework. Through ablation studies, we demonstrate that general-purpose semantic text embeddings are rich and aligned with speech for toxicity classification purposes. Conducting experiments across multiple languages at scale, we show improvements in voice toxicity classification across five languages and different toxicity categories.

Nameer Hirschkind, Xiao Yu, Mahesh Kumar Nandwana et al.

We introduce DiffuseST, a low-latency, direct speech-to-speech translation system capable of preserving the input speaker's voice zero-shot while translating from multiple source languages into English. We experiment with the synthesizer component of the architecture, comparing a Tacotron-based synthesizer to a novel diffusion-based synthesizer. We find the diffusion-based synthesizer to improve MOS and PESQ audio quality metrics by 23\% each and speaker similarity by 5\% while maintaining comparable BLEU scores. Despite having more than double the parameter count, the diffusion synthesizer has lower latency, allowing the entire model to run more than 5$\times$ faster than real-time.