Topic: training data

Adaption Labs has launched Invent a Dataset, a tool that creates training data based on task descriptions without needing a seed corpus or manual labeling, aiming to streamline the training process for machine learning models. This innovation could significantly reduce the time and effort required in preparing datasets, making model development more accessible and efficient.

marktechpost.com↗

Benchmark contamination can inflate scores by leaking test items into training data, but its impact on reordering LLM leaderboards is limited, suggesting the reliability threat may be overstated.

arxiv.org↗

Large Vision-Language Models (LVLMs), while effective in various tasks, can exhibit biased behavior due to social biases in their training data. Researchers propose a method called Counterfactual Ensemble Decoding to mitigate these biases, highlighting the importance of addressing fairness in AI systems.

arxiv.org↗