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

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI 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. 16058
    Prev
    Next
    INDEX
    Explanations

    phrases related to accomplishment or achievement

    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
    608
    +0.14
    0.6%
    373
    +0.11
    0.5%
    665
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    608
    +0.14
    0.02
    665
    +0.11
    0.02
    613
    +0.11
    0.02
    Negative Logits
    soDeliveryDate
    -0.82
    eyes
    -0.72
    Limited
    -0.71
    DOWN
    -0.70
    ese
    -0.70
    Ca
    -0.68
     Context
    -0.66
     Continued
    -0.66
     Compass
    -0.66
    Edited
    -0.65
    POSITIVE LOGITS
     become
    1.10
     marry
    1.03
     earn
    1.03
     achieve
    1.02
     formulate
    1.02
     dismantle
    1.01
     explore
    1.00
     pursue
    1.00
     create
    0.99
     establish
    0.98
    Activations Density 0.038%

    No Known Activations