Replay-Based Continual Adaptation for English-Norwegian Petroleum Machine Translation
Tech Stack
Description
This project studies this question in a source-update setting: an MT model is first adapted to an earlier NPD petroleum corpus and then updated with newer Equinor data. The goal is to compare different update strategies and understand the trade-off between learning from the newer source and retaining performance on the earlier source.
As the project and experimental lead, my contributions included:
- Developing a continual-adaptation pipeline comparing single-source, mixed-source, sequential, and replay-based LoRA training.
- Evaluating new-source adaptation, previous-source retention, and catastrophic forgetting using translation-quality and terminology metrics.
Research Question
How can an MT model learn from a new data source without forgetting what it learned from the previous source?
The answer is evaluated through a controlled comparison of single-source, mixed-source, continual, and replay-based update strategies.
Experimental Design
The experiments compare single-source training, mixed-source training, continual adaptation, and replay-based continual adaptation. All settings use fixed NPD and Equinor test sets. In total, 27 LoRA adapters were trained across three random seeds.
| Strategy | What it tests |
|---|---|
| NPD only | Performance before the model is updated with Equinor data. |
| Equinor only | Performance when the model is trained directly on Equinor data. |
| Mixed training | Performance when all NPD and Equinor data are used together. |
| Continual adaptation | What happens when the NPD model is updated using only Equinor data. |
| Continual + replay | Whether adding a small amount of NPD data during the Equinor update reduces forgetting. |
Results
| Update strategy | NPD BLEU | Equinor BLEU | What it shows |
|---|---|---|---|
| NPD only | 61.16 | 34.03 | Strong on NPD but weak on Equinor. |
| Equinor only | 50.37 | 42.29 | Strong on Equinor but weak on NPD. |
| Mixed training | 59.92 | 42.26 | Strong overall results using all data. |
| Continual adaptation | 54.73 | 42.72 | Learns Equinor well but forgets some NPD knowledge. |
| Continual + 5% replay | 58.21 | 42.11 | Reduces forgetting using a small amount of NPD data. |
| Continual + 10% replay | 59.22 | 41.85 | Gives the best lightweight balance across both sources. |
Main Finding
Replay-based continual adaptation gives the best practical update strategy. Continual adaptation improves performance on newer Equinor data but reduces performance on the earlier NPD source. Adding 5-10% NPD replay recovers much of the NPD loss while preserving most of the Equinor gain.