Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Apollo Research · Taylor · Identifying Functionally Important Features with End-to-End Sparse Dictionary Learning
    3. GPT2-Small
    4. Residual Similar CE E2E Recon
    5. 6-RES_SCEFR-AJT
    6. 13782
    Prev
    Next
    INDEX
    Explanations

    sentences that include the expression "I can say"

    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
    280
    +0.15
    0.5%
    678
    +0.12
    0.4%
    70
    +0.12
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    70
    +0.15
    0.08
    411
    +0.12
    0.05
    195
    +0.12
    0.02
    Negative Logits
    ahime
    -1.01
     captcha
    -0.89
     flo
    -0.87
    egal
    -0.85
    laws
    -0.82
     craving
    -0.81
    awar
    -0.80
    ghan
    -0.80
    ollah
    -0.79
    Balt
    -0.79
    POSITIVE LOGITS
    rist
    0.99
    ivably
    0.99
     confidently
    0.94
    ħĭ
    0.93
     definitively
    0.93
    quist
    0.92
    idate
    0.91
    uine
    0.91
    ĵĺ
    0.88
     unequivocally
    0.88
    Activations Density 0.674%

    No Known Activations