© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Apollo Research · Taylor · Identifying Functionally Important Features with End-to-End Sparse Dictionary Learning
    5. GPT2-Small
    7. Residual Similar CE Local
    9. 6-RES_SCL-AJT
    11. 36602
    Prev
    Next
    INDEX
    Explanations

    financial transactions or thefts

    oai_token-act-pair · gpt-3.5-turbo

    instances of the word "from" particularly in contexts involving taking, stealing, or moving something

    oai_token-act-pair · gpt-3.5-turboTriggered by @danbraun
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    neuronpedia/gpt2-small__res_scl-ajt/6-res_scl-ajt
    Prompts (Dashboard)
    12,288 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    46,080
    Data Type
    torch.float32
    Hook Point
    blocks.6.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    apollo-research/Skylion007-openwebtext-tokenizer-gpt2
    Hook Point Layer
    6
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    447
    +0.25
    1.0%
    393
    +0.12
    0.5%
    608
    +0.12
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    362
    +0.25
    0.19
    385
    +0.12
    0.21
    393
    +0.12
    0.25
    Negative Logits
    impact
    -0.79
    important
    -0.79
    ratulations
    -0.78
    iveness
    -0.78
    achy
    -0.77
    arse
    -0.77
    umbered
    -0.76
    monary
    -0.75
    aqu
    -0.75
    airs
    -0.75
    POSITIVE LOGITS
     afar
    1.29
     abroad
    1.13
     whence
    0.95
     thence
    0.93
     scratch
    0.90
     somewhere
    0.83
     neighbouring
    0.79
     anywhere
    0.77
     elsewhere
    0.76
     overseas
    0.76
    Activations Density 9.414%

    No Known Activations