LingBERT, Linguistic Knowledge Injection into Attention Mechanism Based on a Hybrid Masking Strategy
Bibliographic record
Abstract
In this paper, we propose lingBERT, a transformers based language model. We present two architectures of lingBERT based on hybrid masking strategy. Both models are inspired by BERT base. Our model introduces linguistic knowledge (syntactic dependencies) injection into attention mechanisms. Models like BERT employ random masking of tokens during training, which can lead to inefficiencies in capturing syntactic and semantic dependencies. To address this, our method uses two masking strategies. The first one masks words with syntactic dependencies. The second one uses a low percentage of randomly masked words. Resulted tokens from both strategies are then handed over tow proposed architectures of lingBERT. This strategy ensures that linguistic relationships are preserved and learned more effectively. Additionally, we maintain a low randomness ratio of masked tokens to avoid overfitting and enhance the model generalization. Through comprehensive experiments and evaluations, our approach demonstrates significant improvements in capturing context, leading to better performance across various NLP tasks. Furthermore, our approach affords an interpretability about our model inner workings throughout the learning process. This work offers a new direction toward knowledge injection into attention-mechanism based models, leading to advancing the capabilities of language understanding systems.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".