Does Reinforcement Learning Really Incentivize Reasoning Capacity in LLMs Beyond the Base Model?
Author not recordedDate not recordedOriginal︎ ↗
Examinations
None recorded.
Text
No stored copy. Only the passages claims quote are recorded.
Surprisingly, we find that the current training setup does not elicit fundamentally new reasoning patterns. While RLVR-trained models outperform their base models at small k (e.g., k = 1), the base models achieve a higher pass@k score when k is large. Coverage and perplexity analyses show that the observed reasoning abilities originate from and are bounded by the base model.
Not yet read by this claim’s Steward.
While RLVR-trained models outperform their base models at small k (e.g., k = 1), the base models achieve a higher pass@k score when k is large.
Not yet read by this claim’s Steward.
we critically examine the current state of RLVR by systematically probing the reasoning capability boundaries of RLVR-trained LLMs across various model families, RL algorithms, and math, coding, and visual reasoning benchmarks, using pass@k at large k values as the evaluation metric.
Not yet read by this claim’s Steward.
Lineage across 3 claims
Record
- Kind
- unknown
- Authors
- Not recorded
- Publisher
- Not recorded
- Published
- Not recorded
- DOI
- None
- Link
- https://arxiv.org/abs/2504.13837
- Archived copy
- None found
- Stored text
- None; quoted passages only
- Claims extracted
- Not recorded
- Versions
- One known
- Copies
- None recorded
- Corrections
- Not yet checked
- Facts checked
- Not yet
History 1
- Oct 30, 2026Entered the graphEntered the graph; no copy stored