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

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsBlogSlackPrivacy & 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 Local
    5. 6-RES_SCL-AJT
    6. 40204
    Prev
    Next
    INDEX
    Explanations

    phrases or words related to facilitating ease or comfort

    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
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    447
    +0.17
    0.9%
    195
    +0.14
    0.7%
    606
    +0.09
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    608
    +0.17
    0.19
    393
    +0.14
    0.29
    73
    +0.09
    0.07
    Negative Logits
    ificial
    -0.79
    ertodd
    -0.67
    arians
    -0.65
    ichick
    -0.64
     Canary
    -0.62
    aries
    -0.62
    eters
    -0.60
    arium
    -0.60
    acular
    -0.59
    alion
    -0.59
    POSITIVE LOGITS
    liness
    0.87
     ease
    0.76
    UGE
    0.70
     breeze
    0.68
    ï¸
    0.65
    joy
    0.63
    way
    0.63
    Http
    0.62
    Reviewer
    0.62
     Pradesh
    0.61
    Activations Density 15.093%

    No Known Activations