DeepSeek signals: data quality ROI now beats novel post-training algorithms
A_K_Nain · x · 2026-09-11
A notable signal: DeepSeek — usually first to pioneer novel algorithms and architectures — says the ROI of improving data quality now far exceeds working on novel post-training algorithms.
The commenter argues this has been true for non-lab practitioners for a while: put 80% of your post-training effort into data —
- Hire experts to dig through RL tasks
- Manually sift rollouts and SFT data, remove suspicious samples, verify all tasks are solvable
- Ensure diversity in both difficulty and category
Related event: DeepSeek Says Data Quality Now Beats New Post-Training Algorithms(3 posts)→
More from Models
- Unverified: GPT-6.1 Terra reportedly released, leaked chart shows top score per dollar on Agents' Last Exam — ZeroStateReflex · 2026-10-27
- GPT-6.1 Terra cuts factual error rate from 22.3% to 4.3%, an ~81% drop — haider1 · 2026-10-27
- GPT-6.1 Terra lands: early users praise speed, value and subagent performance — BenBajarin · 2026-10-27
- Leaked GPT-6.1 Terra benchmarks: cheap but 'boring', still far behind Astra and Claude — Angaisb_ · 2026-10-27
- Artificial Analysis Ships Intelligence Index v4.3.2 With GPT-6.1 Terra Comparison — ArtificialAnlys · 2026-10-27
- GPT-6.1 Terra cuts hallucination rate by 50 points while lifting accuracy 2 points — ArtificialAnlys · 2026-10-27