What is the impact of scaling model size (e.g., 3B, 7B, 13B params) on zero-shot cross-lingual performance when using English
Description
Intermediate-task training---fine-tuning a pretrained model on an intermediate task before fine-tuning again on the target task---often improves model performance substantially on language understanding tasks in monolingual English settings. We investigate whether English intermediate-task training is still helpful on non-English target tasks. Using nine intermediate language-understanding tasks, we evaluate intermediate-task transfer in a zero-shot cross-lingual setting on the XTREME benchmark. We see large improvements from intermediate training on the BUCC and Tatoeba sentence retrieval tas
Research goal: What is the impact of scaling model size (e.g., 3B, 7B, 13B params) on zero-shot cross-lingual performance when using English intermediate-task training on XTREME-R, and does this scaling affect inference efficiency differently in quantized vs. full-precision models?
Autonomous synthesis report generated by Assignee Research. Tribunal consensus score: 7.8/10.
Notes
Files
paper.pdf
Files
(77.6 kB)
| Name | Size | Download all |
|---|---|---|
|
md5:2c2996f3d1b8f09b51cec2c938b4bdf3
|
77.6 kB | Preview Download |