Publications

@conference{capone-etal-2025-class,

title = {CLASS-IT: Conversational and Lecture-Aligned Small-Scale Instruction Tuning for BabyLMs},

author = {Luca Capone and Alessandro Bondielli and Alessandro Lenci},

editor = {Lucas Charpentier and Leshem Choshen and Ryan Cotterell and Mustafa Omer Gul and Michael Y. Hu and Jing Liu and Jaap Jumelet and Tal Linzen and Aaron Mueller and Candace Ross and Raj Sanjay Shah and Alex Warstadt and Ethan Gotlieb Wilcox and Adina Williams},

url = {https://aclanthology.org/2025.babylm-main.30/},

doi = {10.18653/v1/2025.babylm-main.30},

isbn = {TODO},

year  = {2025},

date = {2025-11-01},

urldate = {2025-11-01},

booktitle = {Proceedings of the First BabyLM Workshop},

pages = {436–444},

publisher = {Association for Computational Linguistics},

address = {Suzhou, China},

abstract = {This work investigates whether small-scale LMs can benefit from instruction tuning (IT). We compare conversational and question–answering IT datasets, applied either in a merged or sequential curriculum, using decoder-only models with 100M and 140M parameters. Evaluation spans both fine-tuning (SuperGLUE) and zero-shot (BLiMP, EWoK, WUGs, entity tracking, and psycholinguistic correlation) settings. Results show that IT yields small but consistent gains in fine-tuning scenarios, with sequential curricula outperforming merged data; however, improvements do not consistently transfer to zero-shot tasks, suggesting a trade-off between interaction-focused adaptation and broad linguistic generalization. These results highlight both the potential and the constraints of adapting human-inspired learning strategies to low-resource LMs, and point toward hybrid, curriculum-based approaches for enhancing generalization under ecological training limits.},

keywords = {},

pubstate = {published},

tppubtype = {conference}

}

@conference{<LineBreak>	11568_1327955.2,

title = {BAMBI Goes to School: Evaluating Italian BabyLMs with Invalsi-ITA},

author = {Luca Capone and Alice Suozzi and Gianluca E Lebani and Alessandro Lenci},

url = {https://clic2025.unica.it/wp-content/uploads/2025/09/15_main_long.pdf},

year  = {2025},

date = {2025-01-01},

urldate = {2025-01-01},

booktitle = {Proceedings of the Eleventh Italian Conference on Computational Linguistics (CLiC-it 2025)},

pages = {1–12},

publisher = {Università di Cagliari},

keywords = {},

pubstate = {published},

tppubtype = {conference}

}

@conference{<LineBreak>	11568_1327950.2,

title = {BaBIEs: A Benchmark for the Linguistic Evaluation of Italian Baby Language Models},

author = {Luca Capone and Alice Suozzi and Gianluca Lebani and Alessandro Lenci},

url = {https://aclanthology.org/2024.clicit-1.20/},

year  = {2024},

date = {2024-01-01},

urldate = {2024-01-01},

booktitle = {Proceedings of the 10th Italian Conference on Computational Linguistics (CLiC-it 2024)},

pages = {157–170},

publisher = {CEUR Workshop Proceedings},

keywords = {},

pubstate = {published},

tppubtype = {conference}

}

@conference{<LineBreak>	11568_1327944,

title = {ConcreteGPT: A Baby GPT-2 Based on Lexical Concreteness and Curriculum Learning},

author = {L. Capone and A. Bondielli and A. Lenci},

url = {https://aclanthology.org/2024.conll-babylm.16/},

year  = {2024},

date = {2024-01-01},

urldate = {2024-01-01},

booktitle = {The 2nd BabyLM Challenge at the 28th Conference on Computational Natural Language Learning},

pages = {189–196},

keywords = {},

pubstate = {published},

tppubtype = {conference}

}