The Price of Reasoning: Cost-Quality Tradeoffs in Reinforcement Learning for Neural Machine Translation
Merged summary
TL;DR - This paper studies whether explicit reasoning traces improve RLVR-trained neural machine translation and at what computational cost. Reasoning during inference improves translation quality but increases output-token usage.
- Systematically omits reasoning traces during either training or inference to isolate their effects.
- Finds that reasoning is particularly beneficial during inference.
- Evaluates the tradeoff between higher translation quality and increased token-generation costs.
- Examines RLVR for specialized translation settings, including legal documents.
Sources (1)
The Price of Reasoning: Cost-Quality Tradeoffs in Reinforcement Learning for Neural Machine Translation
TL;DR - This paper studies whether explicit reasoning traces improve RLVR-trained neural machine translation and at what computational cost. Reasoning during inference improves translation quality but increases output-token usage.
- Systematically omits reasoning traces during either training or inference to isolate their effects.
- Finds that reasoning is particularly beneficial during inference.
- Evaluates the tradeoff between higher translation quality and increased token-generation costs.
- Examines RLVR for specialized translation settings, including legal documents.