Calibration, Entropy Rates, and Memory in Language Models
By Mark Braverman, Xinyi Chen, Sham M. Kakade, Karthik Narasimhan, Cyril Zhang, Yi Zhang. 2019
50 citations recorded by the source corpus.
Building accurate language models that capture meaningful long-term dependencies is a core challenge in natural language processing. Towards this end, we present a calibration-based approach to measure long-term discrepancies between a generative sequence model and the true distribution, and use these discrepancies to improve the model. Empirically, we show that state-of-the-art language models, including LSTMs and Transformers, are \emph{miscalibrated}: the entropy rates of their generations drift dramatically upward over time. We then provide provable methods to mitigate this phenomenon. Furthermore, we show how this calibration-based approach can also be used to measure the amount of memory that language models use for prediction.
Subfields: Artificial Intelligence, Cultural Studies
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
- Thibault Sellam
- Ankur P. Parikh
- Karthik Narasimhan
- Shunyu Yao
- Jiwei Li
- Fei Wu
- Mohit Bansal
- Theodore R. Sumers
Citations in the corpus
- GPT-NER: Named Entity Recognition via Large Language Models (2023, background): “LLMs significantly suffer from the hallucination or overprediction issue (Braverman et al., 2020; Jiang et al., 2021; Zhao et al., 2021).”
- InfoLM: A New Metric to Evaluate Summarization & Data2Text Generation (2021, background): “g entropy rates, (Braverman et al. 2020) temperature scaling (Platt et al.”
- Sticking to the Facts: Confident Decoding for Faithful Data-to-Text Generation (2019, methodology): “It allows us to weigh more on the confidence of generation, without sacrificing perplexity (Braverman et al., 2019).”
- Cognitive Architectures for Language Agents (2023, background): “On the other hand, LLM development might be influenced or even shaped by the increased usage of reasoning toward complex decision making, and better support decision making by solving known issues such as over-confidence and miscalibration (Jiang et al., 2021; Braverman et al., 2020; Chen et al., 20”