Quiet Feature Learning in Algorithmic Tasks
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866908762755825664 |
|---|---|
| author | Naidu, Prudhviraj Wang, Zixian Bergen, Leon Paturi, Ramamohan |
| author_facet | Naidu, Prudhviraj Wang, Zixian Bergen, Leon Paturi, Ramamohan |
| contents | We train Transformer-based language models on ten foundational algorithmic tasks and observe pronounced phase transitions in their loss curves that deviate from established power-law scaling trends. Over large ranges of compute, the validation loss barely improves, then abruptly decreases. Probing the models' internal representations reveals that quiet features are learned prior to any decrease in task loss. These quiet features represent intermediate algorithmic computations that do not by themselves improve the output loss. Ablation experiments demonstrate that individual quiet features are causally necessary for task performance. Our results demonstrate that substantial representational progress can remain hidden beneath an apparently flat loss curve, challenging the prevailing use of cross-entropy as a proxy for learning and motivating richer diagnostics for monitoring model training. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2505_03997 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Quiet Feature Learning in Algorithmic Tasks Naidu, Prudhviraj Wang, Zixian Bergen, Leon Paturi, Ramamohan Machine Learning Computation and Language We train Transformer-based language models on ten foundational algorithmic tasks and observe pronounced phase transitions in their loss curves that deviate from established power-law scaling trends. Over large ranges of compute, the validation loss barely improves, then abruptly decreases. Probing the models' internal representations reveals that quiet features are learned prior to any decrease in task loss. These quiet features represent intermediate algorithmic computations that do not by themselves improve the output loss. Ablation experiments demonstrate that individual quiet features are causally necessary for task performance. Our results demonstrate that substantial representational progress can remain hidden beneath an apparently flat loss curve, challenging the prevailing use of cross-entropy as a proxy for learning and motivating richer diagnostics for monitoring model training. |
| title | Quiet Feature Learning in Algorithmic Tasks |
| topic | Machine Learning Computation and Language |
| url | https://arxiv.org/abs/2505.03997 |