Introduces LIHA ablation to locate first-token broadcaster heads and provides causal evidence that instruction tuning localizes language identity circuits to early layers in transformers.
Understanding and Mitigating Language Confusion in LLM s
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
fields
cs.CL 2years
2026 2verdicts
UNVERDICTED 2representative citing papers
Broad empirical evaluation finds that fine-tuning heuristics for source-language choice in cross-lingual transfer do not hold reliably under in-context learning.
citing papers explorer
-
First-Token Broadcasters: Mechanistic Origins of Language Identity and Distributed Robustness in Transformers
Introduces LIHA ablation to locate first-token broadcaster heads and provides causal evidence that instruction tuning localizes language identity circuits to early layers in transformers.
-
When English Isn't the Best Teacher: Source Language Effects in Cross-Lingual In-Context Learning
Broad empirical evaluation finds that fine-tuning heuristics for source-language choice in cross-lingual transfer do not hold reliably under in-context learning.