Formality is Favored: Unraveling the Learning Preferences of Large Language Models on Data with Conflicting Knowledge
- URL: http://arxiv.org/abs/2410.04784v1
- Date: Mon, 7 Oct 2024 06:49:41 GMT
- Title: Formality is Favored: Unraveling the Learning Preferences of Large Language Models on Data with Conflicting Knowledge
- Authors: Jiahuan Li, Yiqing Cao, Shujian Huang, Jiajun Chen,
- Abstract summary: Large language models have shown excellent performance on many knowledge-intensive tasks.
However, pretraining data tends to contain misleading and even conflicting information.
This study systematically analyze LLMs' learning preferences for data with conflicting knowledge.
- Score: 55.65162959527848
- License: http://creativecommons.org/licenses/by/4.0/
- Abstract: Having been trained on massive pretraining data, large language models have shown excellent performance on many knowledge-intensive tasks. However, pretraining data tends to contain misleading and even conflicting information, and it is intriguing to understand how LLMs handle these noisy data during training. In this study, we systematically analyze LLMs' learning preferences for data with conflicting knowledge. We find that pretrained LLMs establish learning preferences similar to humans, i.e., preferences towards formal texts and texts with fewer spelling errors, resulting in faster learning and more favorable treatment of knowledge in data with such features when facing conflicts. This finding is generalizable across models and languages and is more evident in larger models. An in-depth analysis reveals that LLMs tend to trust data with features that signify consistency with the majority of data, and it is possible to instill new preferences and erase old ones by manipulating the degree of consistency with the majority data.
Related papers
- Learning with Less: Knowledge Distillation from Large Language Models via Unlabeled Data [54.934578742209716]
In real-world NLP applications, Large Language Models (LLMs) offer promising solutions due to their extensive training on vast datasets.
LLKD is an adaptive sample selection method that incorporates signals from both the teacher and student.
Our comprehensive experiments show that LLKD achieves superior performance across various datasets with higher data efficiency.
arXiv Detail & Related papers (2024-11-12T18:57:59Z) - Dynamic Uncertainty Ranking: Enhancing Retrieval-Augmented In-Context Learning for Long-Tail Knowledge in LLMs [50.29035873837]
Large language models (LLMs) can learn vast amounts of knowledge from diverse domains during pre-training.
Long-tail knowledge from specialized domains is often scarce and underrepresented, rarely appearing in the models' memorization.
We propose a reinforcement learning-based dynamic uncertainty ranking method for ICL that accounts for the varying impact of each retrieved sample on LLM predictions.
arXiv Detail & Related papers (2024-10-31T03:42:17Z) - Accelerating Large Language Model Pretraining via LFR Pedagogy: Learn, Focus, and Review [50.78587571704713]
Learn-Focus-Review (LFR) is a dynamic training approach that adapts to the model's learning progress.
LFR tracks the model's learning performance across data blocks (sequences of tokens) and prioritizes revisiting challenging regions of the dataset.
Compared to baseline models trained on the full datasets, LFR consistently achieved lower perplexity and higher accuracy.
arXiv Detail & Related papers (2024-09-10T00:59:18Z) - Elephants Never Forget: Memorization and Learning of Tabular Data in Large Language Models [21.10890310571397]
Large Language Models (LLMs) can be applied to a diverse set of tasks, but the critical issues of data contamination and memorization are often glossed over.
This work introduces a variety of different techniques to assess whether a language model has seen a dataset during training.
We then compare the few-shot learning performance of LLMs on datasets that were seen during training to the performance on datasets released after training.
arXiv Detail & Related papers (2024-04-09T10:58:21Z) - Dated Data: Tracing Knowledge Cutoffs in Large Language Models [47.987664966633865]
We propose a simple approach to estimate effective cutoffs on the resource-level temporal alignment of an LLM.
We find that effective cutoffs often differ from reported cutoffs.
Our analysis reveals two reasons for these inconsistencies: (1) temporal biases of CommonCrawl data due to non-trivial amounts of old data in new dumps and (2) complications in LLM deduplication schemes involving semantic duplicates and lexical near-duplicates.
arXiv Detail & Related papers (2024-03-19T17:57:58Z) - Learning to Reduce: Optimal Representations of Structured Data in
Prompting Large Language Models [42.16047343029512]
Large Language Models (LLMs) have been widely used as general-purpose AI agents.
We propose a framework, Learning to Reduce, that fine-tunes a language model to generate a reduced version of an input context.
We show that our model achieves comparable accuracies in selecting the relevant evidence from an input context.
arXiv Detail & Related papers (2024-02-22T00:41:23Z) - Supervised Knowledge Makes Large Language Models Better In-context Learners [94.89301696512776]
Large Language Models (LLMs) exhibit emerging in-context learning abilities through prompt engineering.
The challenge of improving the generalizability and factuality of LLMs in natural language understanding and question answering remains under-explored.
We propose a framework that enhances the reliability of LLMs as it: 1) generalizes out-of-distribution data, 2) elucidates how LLMs benefit from discriminative models, and 3) minimizes hallucinations in generative tasks.
arXiv Detail & Related papers (2023-12-26T07:24:46Z) - LLMaAA: Making Large Language Models as Active Annotators [32.57011151031332]
We propose LLMaAA, which takes large language models as annotators and puts them into an active learning loop to determine what to annotate efficiently.
We conduct experiments and analysis on two classic NLP tasks, named entity recognition and relation extraction.
With LLMaAA, task-specific models trained from LLM-generated labels can outperform the teacher within only hundreds of annotated examples.
arXiv Detail & Related papers (2023-10-30T14:54:15Z) - CulturaX: A Cleaned, Enormous, and Multilingual Dataset for Large
Language Models in 167 Languages [86.90220551111096]
Training datasets for large language models (LLMs) are often not fully disclosed.
We present CulturaX, a substantial multilingual dataset with 6.3 trillion tokens in 167 languages.
arXiv Detail & Related papers (2023-09-17T23:49:10Z)
This list is automatically generated from the titles and abstracts of the papers in this site.
This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.