Papers with data composition
Heterogeneity in Formal Linguistic Competence of Language Models: Is Data the Real Bottleneck? (2026.findings-acl)
Copied to clipboard
| Challenge: | Large Language Models (LLMs) exhibit a puzzling disparity in their formal linguistic competence, even after training on trillions of tokens. |
| Approach: | They pre-train Large Language Models on 100M-token corpora and inject a minimal amount of synthetic data targeting specific linguistic phenomena into the model. |
| Outcome: | The proposed intervention significantly improves model performance in 8 out of the 9 worst-performing BLiMP paradigms. |
AfriqueLLM: How Data Mixing and Model Architecture Impact Continued Pre-training for African Languages (2026.acl-long)
Copied to clipboard
Hao Yu, Tianyi Xu, Michael A. Hedderich, Wassim Hamidouche, Syed Waqas Zamir, David Ifeoluwa Adelani
| Challenge: | Continued pretraining (CPT) is a practical route to language adaptation, but improvements on demanding capabilities such as mathematical reasoning are limited. |
| Approach: | They propose to use CPT to adapt large language models to African languages . they use math, code, and synthetic translated data to analyze their models . |
| Outcome: | The proposed models improve on multilingual benchmarks and document-level translation. |
From 128K to 4M: Efficient Training of Ultra-Long Context Large Language Models (2026.findings-acl)
Copied to clipboard
| Challenge: | Long-context capabilities are essential for document and video understanding, in-contact learning, and inference-time scaling. |
| Approach: | They propose an efficient training recipe for building ultra-long context LLMs from aligned instruct model, pushing the boundaries of context lengths from 128K to 1M, 2M, and 4M tokens. |
| Outcome: | The proposed model extends the context window while maintaining short context capabilities while maintaining the performance of the existing model. |
A Data-Centric Approach to Generalizable Speech Deepfake Detection (2026.acl-long)
Copied to clipboard
| Challenge: | Speech deepfake detection (SDD) is a critical research area as speech synthesis technologies become more sophisticated. |
| Approach: | They propose a data-centric approach to generalize SDD data from two perspectives . they propose naive aggregation strategies for mixing heterogeneous data and diversity-optimized sampling strategy for a single dataset and multiple datasets. |
| Outcome: | The proposed approach outperforms the naive aggregation baseline on a 12k-hour data pool while using only 3% of the total available data. |
Not-Just-Scaling Laws: Towards a Better Understanding of the Downstream Impact of Language Model Design Decisions (2025.emnlp-main)
Copied to clipboard
Emmy Liu, Amanda Bertsch, Lintang Sutawika, Lindia Tjuatja, Patrick Fernandes, Lara Marinov, Michael Chen, Shreya Singhal, Carolin Lawrence, Aditi Raghunathan, Kiril Gashteovski, Graham Neubig
| Challenge: | Language model performance is largely dependent on pretraining decisions, but scaling laws based on only these two aspects do not always explain downstream task performance. |
| Approach: | They meta-analyze 92 open-source pretrained models to quantify their impact on performance. |
| Outcome: | The framework lays a foundation for more systematic investigation of how model development choices shape final capabilities. |
SciPedia: Unlocking the Value of Scientific Data for Pre-training (2026.acl-long)
Copied to clipboard
| Challenge: | High-quality scientific data is critical for advancing LLMs, yet academic literature remains underutilized. |
| Approach: | They construct a large-scale raw scientific corpus but identify a critical Learnability Gap . they develop a multi-stage pipeline featuring content cleaning and pedagogical augmentation . |
| Outcome: | The proposed approach boosts average performance by +2.12 (3B) and +2.95 (7B) on in-domain tasks. |