Evolution strategies can full-parameter fine-tune billion-parameter LLMs, outperforming PPO and GRPO on the Countdown task and reward robustness in a conciseness task.
Intrinsic dimensionality explains the effectiveness of language model fine-tuning
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
-
Evolution Strategies at Scale: LLM Fine-Tuning Beyond Reinforcement Learning
Evolution strategies can full-parameter fine-tune billion-parameter LLMs, outperforming PPO and GRPO on the Countdown task and reward robustness in a conciseness task.