method
active
method:k-sparse-probingk-Sparse Probing
Concept detection metric using top-k latents and linear probe.
Neighborhood — ranked by edge-count
Papers (1)
paper
Frameworks (1)
framework
- SAEBenchusesEvaluation suite of eight metrics across four capabilities.
Datasets (5)
dataset
- Amazon ReviewsusesUsed for probing, SCR, TPP.
- Bias in BiosusesUsed for sparse probing, SCR, TPP.
- AG NewsusesTopic probing.
- EuroparlusesLanguage ID probing.
- Programming language probing.
Methods (2)
method
- Sparse Probingrelated_toMethod from Gurnee et al. 2023 for finding feature directions including individual neuron analysis
- Feature selection for sparse probing.
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.
- Standard interpretability technique used to read feature beliefs.
- Probing limited.
- Probing approach avoiding supervision to sidestep complexity-accuracy tradeoff
- Top-down interpretability approach studying linguistic properties at various residual stream stages; contrasted with the paper's bottom-up mechanistic approach
- Used to compute J-space components and occupancy.
- Technique of reading out model beliefs from internal activations before the final answer token is generated
- Attention probe predicting whether an Entity is hallucinated; AUC .94 on validation.
- Interpretability tools that decode information from internal model activations; here, linear probes are used for data attribution.