© 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. 18900
    Prev
    Next
    INDEX
    Explanations

    words related to technological concepts or instructions

    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
    569
    +0.10
    0.5%
    587
    +0.10
    0.5%
    276
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    590
    +0.10
    0.01
    623
    +0.10
    0.01
    288
    +0.10
    0.02
    Negative Logits
    atche
    -0.81
    apesh
    -0.78
    lessness
    -0.76
    abiding
    -0.76
    ieving
    -0.74
    sein
    -0.73
    uay
    -0.73
    ichick
    -0.73
    \\\\\\\\
    -0.72
    fighting
    -0.72
    POSITIVE LOGITS
     extras
    1.01
     accessory
    0.95
     bonus
    0.84
     tang
    0.79
     accompan
    0.78
     hitch
    0.77
     consolation
    0.77
     accessories
    0.77
     extra
    0.74
     supplemental
    0.74
    Activations Density 0.048%

    No Known Activations