Transformer-Squared selectively rescales singular values of an LLM's weights with RL-trained expert vectors, then mixes these experts at inference to adapt to unseen tasks.
Fine-tuning llms with singular value decomposition
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Transformer-Squared: Self-adaptive LLMs
Transformer-Squared selectively rescales singular values of an LLM's weights with RL-trained expert vectors, then mixes these experts at inference to adapt to unseen tasks.