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

    Neuronpedia

    Natural Language
    Autoencoders
    NEW
    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    2. Joseph Bloom · Feature Splitting for GPT2-Small
    3. GPT2-Small
    4. Residual Feature Splitting 24576
    5. 8-RES_FS24576-JB
    6. 8806
    Prev
    Next
    INDEX
    Explanations

    phrases indicating a speaker's role or identity.

    oai_token-act-pair · gemini-1.5-flashTriggered by @johnny

    statements related to societal issues, activism, and calls for unity and action

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs24576-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    24,576
    Data Type
    torch.float32
    Hook Point
    blocks.8.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    8
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    715
    +0.10
    0.5%
    457
    +0.10
    0.4%
    481
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    43
    +0.10
    0.02
    457
    +0.10
    0.02
    546
    +0.10
    0.02
    Negative Logits
    »Ĵ
    -0.73
    OND
    -0.65
    ONSORED
    -0.63
    Stars
    -0.63
    TERN
    -0.62
     eruption
    -0.60
    stage
    -0.59
    usterity
    -0.59
    undown
    -0.59
    usters
    -0.58
    POSITIVE LOGITS
     myself
    0.75
     huh
    0.71
     however
    0.70
     we
    0.68
     please
    0.66
     naturally
    0.64
     she
    0.64
     moreover
    0.64
     sometimes
    0.63
     beware
    0.63
    Activations Density 0.112%

    No Known Activations