---
license: mit
---

<h2 align="center" style="font-size: 2.5em; font-weight: bold; color: #2c3e50;">
  CHRONOBERG: Capturing Language Evolution and Temporal Awareness in Foundation Models
</h2>

<p align="center">
  <a href="https://huggingface.co/datasets/spaul25/Chronoberg" style="margin: 0 10px;">🤗 Dataset</a> |
  <a href="https://github.com/paulsubarna/Chronoberg" style="margin: 0 10px;">🐙 GitHub</a>
  <a href="https://arxiv.org/abs/2509.22360" style="margin: 0 10px;">📖 Arxiv</a>
</p>


We introduce CHRONOBERG, a temporally structured corpus of English book texts spanning 250 years, curated from Project Gutenberg and enriched with a variety of temporal annotations. We also introduce historically calibrated affective Valence-Arousal-Dominance (VAD)  lexicons to support temporally grounded interpretation. With the lexicons at hand, we demonstrate a need for modern LLM-based tools to better situate their detection of discriminatory language and contextualization of sentiment across various time-periods. In fact, we show how language models trained sequentially on CHRONOBERG struggle to encode diachronic shifts in meaning, emphasizing the need for temporally aware training and evaluation pipelines, and positioning CHRONOBERG as a scalable resource for the study of linguistic change and temporal generalization. Disclaimer: This repository and dataset includes language and display of samples that could be offensive to readers. 

## Dataset
Dataset Catalog:
- [x] ChronoBerg [Raw](https://huggingface.co/datasets/spaul25/ChronoBerg/tree/main/dataset): Raw literary text files grouped by their publication year
- [x] ChronoBerg [Pre-processed](https://huggingface.co/datasets/spaul25/ChronoBerg/tree/main/dataset): Preprocessed sentence-splitted sentences grouped by their publication years
- [x] ChronoBerg [Annotated](https://huggingface.co/datasets/spaul25/ChronoBerg/tree/main/dataset): sentence-level valence annotated (for each time interval: 50 year span)
- [x] [Valence Lexicons](https://huggingface.co/datasets/spaul25/Chronoberg/tree/main/Lexicons)
- [x] [Dominance Lexicons](https://huggingface.co/datasets/spaul25/Chronoberg/tree/main/Lexicons)
- [x] [Arousal Lexicons](https://huggingface.co/datasets/spaul25/Chronoberg/tree/main/Lexicons)


## Load Dataset
``` 
from dataset import load_dataset

Chronoberg_raw = load_dataset("spaul25/Chronoberg", data_files="dataset/Chronoberg_raw.jsonl") ## Raw
Chronoberg_preprocessed = load_dataset("spaul25/Chronoberg", data_files="dataset/Chronoberg_preprocessed.jsonl") ## Pre-processed
Chronoberg_annotated = load_dataset("spaul25/Chronoberg", data_files="dataset/Chronoberg_annotated.jsonl") ## Annotated

```

**Pretrained Checkpoints** : 
To construct VAD lexicons on your own, we have also made available the pretrained Word2vec models on the entire dataset and the time-interval-specific slices (50 year intervals) of the dataset. 
Model-Type | 1750-99 | 1800-49 | 1850-99 | 1900-49 | 1950-99 |
--- | :---: | :---: | :---: |:---: |:---:
word2vec | [word2vec_1750](https://huggingface.co/datasets/spaul25/ChronoBerg/blob/main/pretrained_models/word2vec_interval_1750.model) | [word2vec_1800](https://huggingface.co/datasets/spaul25/ChronoBerg/blob/main/pretrained_models/word2vec_interval_1800.model) | [word2vec_1850](https://huggingface.co/datasets/spaul25/ChronoBerg/blob/main/pretrained_models/word2vec_interval_1850.model) | [word2vec_1900](https://huggingface.co/datasets/spaul25/ChronoBerg/blob/main/pretrained_models/word2vec_interval_1900.model) | [word2vec_1950](https://huggingface.co/datasets/spaul25/ChronoBerg/blob/main/pretrained_models/word2vec_interval_1950.model) |

**Recommended Dataset Splits**

We have also made available the training and test sets to reproduce the LLM experiments in our [paper](https://arxiv.org/abs/2509.22360). More ways to produce train and tests can be found in our [github](https://github.com/paulsubarna/Chronoberg)

**Main Results**
Here are a few of the main results from our paper. 

A comparison of all continual learning strategies used to train an LLM model sequentially on ChronoBerg can be found below: 

Method | Perplexity | Forward Gen. | Best Case | Worst Case 
--- | :---: | :---: | :---: |:---:
Sequential FT | 34\% &uarr; | 33\% &uarr; | 4.58 (1750--99) | 6.64 (1950--2000) 
EWC           | 12\% &uarr; | 29\% &uarr; | 4.65 (1800--49) | 6.77 (1950--2000) 
LoRA          | 15\% &uarr; | 27\% &uarr; | 4.48 (1850--99) | 6.19 (1950--2000) 

#### Lexical Analysis
We have used our lexicons to analyze words that have undergone shifts from being positive to negative or negative to positive. Here are few instances of such words. 

![Lexical](figures/Lexical_Analysis.png)

**How to cite us**
 ```bibtex
@misc{hegde2025chronobergcapturinglanguageevolution,
      title={CHRONOBERG: Capturing Language Evolution and Temporal Awareness in Foundation Models}, 
      author={Niharika Hegde and Subarnaduti Paul and Lars Joel-Frey and Manuel Brack and Kristian Kersting and Martin Mundt and Patrick Schramowski},
      year={2025},
      eprint={2509.22360},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2509.22360}, 
}
 ```