© 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 768
    9. 8-RES_FS768-JB
    11. 669
    Prev
    Next
    INDEX
    Explanations

    words related to political discussions and geographical locations

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs768-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    768
    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
    649
    +0.11
    0.5%
    548
    +0.10
    0.5%
    542
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    288
    +0.11
    0.32
    575
    +0.10
    0.24
    325
    +0.10
    0.18
    Negative Logits
    ificantly
    -0.70
    ameless
    -0.68
    okingly
    -0.68
    arently
    -0.65
    ModLoader
    -0.64
    untarily
    -0.63
    appropriately
    -0.62
    istered
    -0.62
    ateful
    -0.60
    ependent
    -0.59
    POSITIVE LOGITS
     imaginable
    0.91
     (>
    0.87
    smanship
    0.87
    vironments
    0.84
     (<
    0.77
     awaits
    0.76
    nings
    0.69
    ainment
    0.68
    ounters
    0.68
    izen
    0.67
    Activations Density 8.629%

    No Known Activations