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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    1. Home
    3. Joseph Bloom · Feature Splitting for GPT2-Small
    5. GPT2-Small
    7. Residual Feature Splitting 24576
    9. 8-RES_FS24576-JB
    11. 7219
    Prev
    Next
    INDEX
    Explanations

    phrases related to unlocking new features, abilities, or content in a game

    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
    496
    +0.14
    0.6%
    288
    +0.11
    0.5%
    343
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    326
    +0.14
    0.01
    496
    +0.11
    0.01
    688
    +0.10
    0.01
    Negative Logits
     Drawn
    -0.69
    ogram
    -0.66
    geist
    -0.65
    cas
    -0.64
    ograms
    -0.63
    ochemical
    -0.62
    umbn
    -0.62
    lli
    -0.62
    ospital
    -0.62
    cats
    -0.61
    POSITIVE LOGITS
     unlocks
    0.86
     unlocking
    0.85
     Unlock
    0.84
     unlock
    0.82
     unlocked
    0.81
    ments
    0.79
    locking
    0.76
    locked
    0.75
     recess
    0.73
    ibility
    0.72
    Activations Density 0.019%

    No Known Activations