© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW 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. 31233
    Prev
    Next
    INDEX
    Explanations

    phrases related to legal proceedings and criminal activity, particularly involving prosecutors and charges

    oai_token-act-pair · gpt-3.5-turbo
    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
    Neuron Alignment
    Index
    Value
    % of L₁
    447
    +0.26
    1.2%
    571
    +0.13
    0.6%
    288
    +0.13
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    679
    +0.26
    0.18
    145
    +0.13
    0.07
    618
    +0.13
    0.12
    Negative Logits
    oken
    -0.76
    yip
    -0.71
    avery
    -0.69
    lo
    -0.69
    opter
    -0.68
    eem
    -0.67
    roots
    -0.66
    opian
    -0.66
    lihood
    -0.65
    wo
    -0.65
    POSITIVE LOGITS
     prosecutor
    0.90
     indicted
    0.80
     allege
    0.79
     prosecuted
    0.78
     prosecutors
    0.75
     accuse
    0.74
     accuses
    0.73
     investigating
    0.72
     investigator
    0.72
     proced
    0.71
    Activations Density 8.061%

    No Known Activations