Memory-Gym-trained Researcher behavior transfers to LongCodeQA without code-domain training or adaptation: SFT raises accuracy in every context subset, and subsequent Hint-guided GRPO raises it further.

Memory-Gym-trained Researcher behavior transfers to LongCodeQA without code-domain training or adaptation: SFT raises accuracy in every context subset, and subsequent Hint-guided GRPO raises it further. · CiteArk

Measurement results

0/12 assessed

0 supported0 disputed or conflicting12 not assessed

The complete result group stays with its claim. Assessed rows come first; their judgments do not replace the whole-claim conclusion. Expand a row for conditions, sources, and evidence history.

Each row is a measurement with its reported value, latest available observation, and cumulative evidence state.
Metric and conditions	Reported	Latest observation	Evidence judgment
contextSubset: 64K · trainingStage: Untrained	56.58%	—	Not assessed
contextSubset: 128K · trainingStage: Untrained	55.43%	—	Not assessed
contextSubset: 256K · trainingStage: Untrained	49.23%	—	Not assessed
contextSubset: Overall · trainingStage: Untrained	54.08%	—	Not assessed
contextSubset: 64K · trainingStage: SFT only	68.42%	—	Not assessed
contextSubset: 128K · trainingStage: SFT only	64.13%	—	Not assessed
contextSubset: 256K · trainingStage: SFT only	64.62%	—	Not assessed
contextSubset: Overall · trainingStage: SFT only	65.67%	—	Not assessed
contextSubset: 64K · trainingStage: SFT+RL	76.32%	—	Not assessed
contextSubset: 128K · trainingStage: SFT+RL	76.09%	—	Not assessed
contextSubset: 256K · trainingStage: SFT+RL	73.85%	—	Not assessed
contextSubset: Overall · trainingStage: SFT+RL	75.54%	—	Not assessed

Memory-Gym-trained Researcher behavior transfers to LongCodeQA without code-domain training or adaptation: SFT raises accuracy in every context subset, and subsequent Hint-guided GRPO raises it further.

Reported and observed measurements

Assessments (0)

Runs (0)

Measurement results