mlsanityData to 14 Oct 2026

Tracking Universal Features Through Fine-Tuning and Model Merging

arXiv 2410.12391 ↗First posted 16 Oct 2024cs.CL cs.LGcited by 0

Not on the map. Why?

Abstract (arXiv)

We study how features emerge, disappear, and persist across models fine-tuned on different domains of text. More specifically, we start from a base one-layer Transformer language model that is trained on a combination of the BabyLM corpus, and a collection of Python code from The Stack. This base model is adapted to two new domains of text: TinyStories, and the Lua programming language, respectively; and then these two models are merged using these two models using spherical linear interpolation. Our exploration aims to provide deeper insights into the stability and transformation of features across typical transfer-learning scenarios using small-scale models and sparse auto-encoders.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 14 of its 21 references that are in the map. First posted on arXiv 16 Oct 2024.

Builds on

14 of 21 references.

References · 14

  1. 2014Adam: A Method for Stochastic Optimizationcited by 37,240
  2. 2020The Pile: An 800GB Dataset of Diverse Text for Language Modelingcited by 2,411
  3. 2020Don't Stop Pretraining: Adapt Language Models to Domains and Taskscited by 1,565
  4. 2023Sparse Autoencoders Find Highly Interpretable Features in Language Modelscited by 1,420
  5. 2022Model soups: averaging weights of multiple fine-tuned models improves accuracy without increasing inference timecited by 1,396
  6. 2020Beyond Accuracy: Behavioral Testing of NLP models with CheckListcited by 1,006
  7. 2017Sigmoid-Weighted Linear Units for Neural Network Function Approximation in Reinforcement Learningcited by 950
  8. 2019Linear Mode Connectivity and the Lottery Ticket Hypothesiscited by 637
  9. 2023TinyStories: How Small Can Language Models Be and Still Speak Coherent English?cited by 449
  10. 2022The Stack: 3 TB of permissively licensed source codecited by 359
  11. 2015Convergent Learning: Do different neural networks learn the same representations?cited by 220
  12. 2023A Toy Model of Universality: Reverse Engineering How Networks Learn Group Operationscited by 144
2 more
  1. 2022Fusing finetuned models for better pretrainingcited by 110
  2. 2023Call for Papers -- The BabyLM Challenge: Sample-efficient pretraining on a developmentally plausible corpuscited by 62

Built on by

Not cited yet.