How novel are texts generated by language models (LMs) relative to their
training corpora? In this work, we investigate the extent to which modern LMs
generate $n$-grams from their training data, evaluating both (i) the
probability LMs assign to complete training $n$-grams and (ii) $n$-novelty, the
proportion of $n$-grams generated by an LM that did not appear in the training
data (for arbitrarily large $n$). To enable arbitrary-length $n$-gram search
over a corpus in constant time, we develop Rusty-DAWG, a novel search tool
inspired by indexing of genomic data. We compare the novelty of LM-generated
text to human-written text and explore factors that affect generation novelty,
focusing on the Pythia models. We find that, for $n > 4$, LM-generated text is
less novel than human-written text, though it is more novel for smaller $n$.
Larger LMs and more constrained decoding strategies both decrease novelty.
Finally, we show that LMs complete $n$-grams with lower loss if they are less
frequent in the training data. Overall, our results reveal factors influencing
the novelty of LM-generated text, and we release Rusty-DAWG to facilitate
further pretraining data research.

研究了语言模型生成的文本与其训练语料库之间的新颖程度，并通过评估生成的 $n$-gram 的概率和 $n$- 新颖性来揭示影响生成新颖性的因素。发现对于较大的 $n$，语言模型生成的文本不如人为编写的文本新颖，但对于较小的 $n$，则更为新颖。同时发现较大的语言模型和受限的解码策略均会降低新颖性，而低频 $n$-gram 在训练数据中出现的频率越低，语言模型完成它们时的误差也越小。通过发布 Rusty-DAWG 工具以促进进一步的预训练数据研究。

使用 Rusty-DAWG 评估语言模型的 n-Gram 新颖性

Evaluating $n$-Gram Novelty of Language Models Using Rusty-DAWG

Natural language processing models learn word representations based on the
distributional hypothesis, which asserts that word context (e.g.,
co-occurrence) correlates with meaning. We propose that $n$-grams composed of
random character sequences, or $garble$, provide a novel context for studying
word meaning both within and beyond extant language. In particular, randomly
generated character $n$-grams lack meaning but contain primitive information
based on the distribution of characters they contain. By studying the
embeddings of a large corpus of garble, extant language, and pseudowords using
CharacterBERT, we identify an axis in the model's high-dimensional embedding
space that separates these classes of $n$-grams. Furthermore, we show that this
axis relates to structure within extant language, including word
part-of-speech, morphology, and concept concreteness. Thus, in contrast to
studies that are mainly limited to extant language, our work reveals that
meaning and primitive information are intrinsically linked.

本文提出使用 $garble$，即由随机字符序列组成的 $n$-grams，作为语境来研究字词含义，同时探讨了 $garble$ 字向量、终极语言和伪词的嵌入，并通过 CharacterBERT 模型识别嵌入空间中将这些 $n$-grams 进行分离的轴，进而发现含义和原始信息之间的内在联系。

信号在噪音中：使用字符感知语言模型探索随机字符序列中的编码含义

Signal in Noise: Exploring Meaning Encoded in Random Character Sequences with Character-Aware Language Models

We describe a new method for visualizing topics, the distributions over terms
that are automatically extracted from large text corpora using latent variable
models. Our method finds significant $n$-grams related to a topic, which are
then used to help understand and interpret the underlying distribution.
Compared with the usual visualization, which simply lists the most probable
topical terms, the multi-word expressions provide a better intuitive impression
for what a topic is "about." Our approach is based on a language model of
arbitrary length expressions, for which we develop a new methodology based on
nested permutation tests to find significant phrases. We show that this method
outperforms the more standard use of $\chi^2$ and likelihood ratio tests. We
illustrate the topic presentations on corpora of scientific abstracts and news
articles.

提出了一种用于主题可视化的新方法，使用潜在变量模型自动提取大型文本语料库中的 term 分布，并利用与主题相关的重要短语来解释该分布。该方法使用基于任意长度表达式的语言模型，并提出了一种基于嵌套排列检验的新方法来查找重要短语。结果表明，该方法优于标准的卡方检验和似然比检验。应用在科学摘要和新闻文章的语料库上，演示了主题的呈现。