Our work “Can LLMs Really Learn to Translate a Low-Resource Language from One Grammar Book?” is now on arXiv! - in collaboration with @davidstap, @diwuNLP, @c_monz , and Khalil Sima'an from @illc_amsterdam and @ltl_uva 🧵

@SimonHiaubeng @ElliotMurphy91 The principle Maximise Minimal Means is part of one version of minimalist theory. But it's not UG - it's a third factor, domain-general constraint

@Linguist_UR @ElliotMurphy91 Merge, maybe Agree, maybe Labeling. Though I believe there's work ongoing to attribute Merge itself to third factor, domain-general constraints

@RaphaelMerx I'm a fan of this paper! We'd expect exactly the same for Kalamang (if we could collect an OOD test set). In the appendix we show too that the 100-example test set consists of short, easy sentences so a ChrF++ of ~30 is really not that proficient

@ylecun And do not confuse learning from grammar with learning from parallel sentences!

RT @diwuNLP: We show that a grammar book provides little or even no help for translation in LLMs, questioning the recent "truly zero-shot t…

@JeffDean (Plus, Kalamang parallel data has been online since November 2020!)

@jxmnop It turns out LLMs learn most or all translation ability from parallel sentences in the book, not the grammar. And fine-tuning a small translation model matches or beats long-context LLM results! (plus Kalamang parallel data has been online since Nov 2020)

More generally, we suggest that data collection efforts for multilingual XLR tasks like translation are best focused on parallel data over linguistic description, given the advantages in computational cost, token efficiency, availability!

Just returned from MT Marathon 2024 in Prague - thanks to @ufal_cuni for organising a great week! Between the insightful talks and collaboration on a mini research project, I presented a poster of my recent work. And of course, we explored the sights of Prague too - in 30°C heat!

RT @ufal_cuni: The MT Marathon continues on its third day! We already had great talks by @OndrejBojar, @prajdabre1, @zouharvi, and @esalesk…

Thanks to everyone who stopped by at the #EACL2024 SRW poster session - some great discussions and feedback!

I'll be presenting a poster at #EACL2024 today on my work with @RABawden exploring topic-guided example selection for LLM-based translation. Come along at 2pm or read more here!