Study infers Claude protein-sequence training from continuation tests
What happened
On November 18, 2026, the Hacker News front page covered Tu Xinming and Leo Tianlai Chen's research on Claude's protein-sequence abilities. The work used protein continuation tests and concluded from the results that Claude Fable 5.2 and Claude Opus 5.5 were trained on large-scale protein sequences. The report offers the researchers' inference from continuation tests only. It gives no direct disclosure of training data and does not state the test scale, the specific results or when training took place.
Written by AI from the coverage · updated 2 days ago
Coverage
Follow the reports to see the story from different sides.
- Hacker News front pageClaude was trained on protein sequences at scale
Tu Xinming 与 Leo Tianlai Chen 的研究根据蛋白质续写测试,推断 Claude Fable 5.2 和 Claude Opus 5.5 接受过大规模蛋白质序列训练。
Heat over time
Heat now 3·Comparable peak 9(Nov 18 16:00)·Comparable change over 24 hours -49%
The trend only compares accounts observed without gaps, so its range may be smaller than the current heat. Hover or tap the chart for each hour; the left and right arrow keys step through it.