Why did mass-mean probing with cities+neg_cities training data perform poorly for the 70B model, despite larger_than+smaller_than performing well?

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

• Why did mass-mean probing with cities+neg_cities perform poorly for the 70B model, despite mass-mean probing with larger_than+smaller_than performing well?question0.958
  Unexplained result pointing to asymmetry in how training on opposites affects truth probes at 70B scale
• MM probes trained on larger_than+smaller_than achieve lower NIE than those trained on cities+neg_cities despite higher classification accuracy on sp_en_transfinding0.841
  Dissociation between classification accuracy and causal implication; training on opposites does not always help causally
• Why were interventions with mass-mean probe directions extracted from the likely dataset so effective, despite these probes not being accurate at classifying true/false statements?question0.800
  Open question raised in §7.1 about an unexplained anomalous result
• Probes trained on the likely dataset perform worse than chance on datasets with anti-correlations between text probability and truthfinding0.777
  Shows that truth representations are not reducible to text probability representations
• MM probe trained on likely dataset achieves NIE of 0.70 (false→true) on LLaMA-2-13B, surprisingly strong but weaker than truth probesfinding0.770
  Likely-trained MM probe is a surprisingly effective causal baseline due to correlation between truth and probability on sp_en_trans
• There are fewer representations competent for N tasks than M<N tasks, so training more general models should yield fewer possible solutionshypothesis0.768
  Selective pressure toward convergence via task generality
• Mass-mean probe directions outperform LR and CCS in causal intervention experiments (NIE) in 7/8 experimental conditionsfinding0.767
  Core result showing MM is superior to LR for causal implication despite similar classification accuracy
• Training on cities+neg_cities improves OOD generalization, especially on neg_sp_en_transfinding0.766
  Training on statements and their negations mitigates non-truth feature interference in probe directions