© 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 E2E Recon
    9. 6-RES_SCEFR-AJT
    11. 11457
    Prev
    Next
    INDEX
    Explanations

    special characters and symbols, possibly looking for specific patterns or encoding schemes

    oai_token-act-pair · gpt-3.5-turbo
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    neuronpedia/gpt2-small__res_scefr-ajt/6-res_scefr-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
    526
    +0.17
    0.6%
    146
    +0.14
    0.5%
    119
    +0.14
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    659
    +0.17
    0.05
    585
    +0.14
    0.05
    190
    +0.14
    0.04
    Negative Logits
     sucker
    -0.94
     starvation
    -0.88
     provocation
    -0.86
     payoff
    -0.86
     privately
    -0.84
     quo
    -0.79
    pmwiki
    -0.78
    aminer
    -0.78
     leaks
    -0.77
     inference
    -0.75
    POSITIVE LOGITS
    bsite
    1.40
    aving
    1.31
    hold
    1.28
    ez
    1.17
    isconsin
    1.14
    elcome
    1.11
    hei
    1.09
    HI
    1.09
    soType
    1.07
    ï¸
    1.06
    Activations Density 0.787%

    No Known Activations