© 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. 22968
    Prev
    Next
    INDEX
    Explanations

    instances of the word "rop" in various contexts

    oai_token-act-pair · gpt-4o-miniTriggered 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
    461
    +0.12
    0.5%
    7
    +0.11
    0.5%
    466
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    701
    +0.12
    0.01
    466
    +0.11
    0.01
    7
    +0.10
    0.01
    Negative Logits
     diligence
    -0.67
     eminent
    -0.64
     Sins
    -0.59
     passing
    -0.58
    pling
    -0.58
     capsule
    -0.57
     defensively
    -0.57
     resemblance
    -0.56
    auga
    -0.56
    hart
    -0.56
    POSITIVE LOGITS
    oulos
    1.40
    hedon
    1.24
    onent
    1.06
    olitics
    1.03
    rop
    1.02
    ython
    1.00
    tera
    0.98
    edia
    0.97
    ract
    0.96
    ower
    0.95
    Activations Density 0.022%

    No Known Activations