Topic
synthetic data
MedSynth Dataset Offers 10,000 Synthetic Medical Dialogue-Note Pairs to Advance AI Documentation
MedSynth is a novel dataset of synthetic medical dialogues and notes designed to advance Dialogue-to-Note and Note-to-Dialogue tasks. It includes over 10,000 pairs covering 2000+ ICD-10 codes, addressing the scarcity of open-access, privacy-compliant training data.
ArtBoost: Synthetic Data Augmentation Boosts Acoustic-to-Articulatory Inversion with Limited Real Data
A new data augmentation strategy called ArtBoost leverages large-scale speech-mesh datasets from 3D facial animation to improve acoustic-to-articulatory inversion (AAI) models under limited EMA supervision. The method extracts pseudo articulatory trajectories from facial anchors and pre-trains models before fine-tuning on real data, yielding consistent gains in PCC and RMSE across architectures.
Fine-Tuning a 7B Advisor on Free-Tier GPUs: Adapter-Handoff Recipe Published with Synthetic Data Reliability Warning
A new paper from Md Millat Hosen presents a method to fine-tune Mistral-7B-Instruct on free Kaggle/Colab GPUs using QLoRA adapter handoff. The evaluation reveals that while the fine-tuned model better matched synthetic training data, it performed worse on advising quality and factuality compared to the base model, with errors traced to the synthetic data pipeline.
SpecAlign Framework Uses Synthetic Data to Align Large Language Models with Specific Policies
A research paper introduces SpecAlign, a framework that generates synthetic training data from provider-authored model specifications to align large language models with specific policies. The method combines structured rule annotation, controllable instantiation, and multi-agent adversarial data synthesis to create preference pairs for fine-tuning. Experiments show improved rule compliance without sacrificing general capabilities.
New Auditing Framework Detects Synthetic Data Privacy Leaks Without Model Access
A new causal framework for auditing synthetic data detects privacy leaks by distinguishing true disclosures from phantom ones. It uses statistical hypothesis testing with holdout sets, requires no model access or canary insertion, and is orders of magnitude more efficient than shadow-model approaches.
StateGen Platform Generates Synthetic Training Data for Tool-Augmented LLMs with 9.66/10 Hallucination Score
Researchers introduce StateGen, a synthetic data generation platform that produces scored, reasoning-trace-rich training conversations for tool-augmented LLMs. The platform uses a four-role LLM loop and an authoritative state manager to eliminate tool-call hallucinations, achieving a 9.66/10 score across 64,698 evaluated conversations.