finding
active
finding:physics-sycophancy-user-praise-rises-from-1-32-sft-to-2-18-dpo-reward-shaping-4-reaches-1-80Physics sycophancy user-praise rises from 1.32 (SFT) to 2.18 (DPO); reward shaping α=4 reaches 1.80
Table 11.
Source paper
extracted_from(2026) · Leon Bergen · Usha Bhalla · Sidharth Baskaran · Max Loeffler +13
Neighborhood — ranked by edge-count
Papers (1)
paper
Claims (1)
claim
- Prompt-conditioned behaviors are entangled with broader response concepts and not cleanly separablesupportsExplains partial recovery.
Related by similarity (8)
cosine ≥ 0.65 · no typed edgeEntities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.
- Fig. 12.
- Benchmark identification rate rises from 0.08 (SFT) to 0.25 (DPO); interventions barely reduce itfinding0.790Table 11.
- Clamping sycophantic praise feature 1M/847723 to 5x max activation causes over-the-top praise.finding0.787Demonstrates causal role in sycophancy.
- SAE analysis shows sycophancy is primarily stylistic rather than content-based
- Table 11.
- Validates robustness of alignment metric choice
- Fig. 10, Table 10.
- Table 2, row 3, showing equivalence when prior preferences match rewards.