© 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. 19304
    Prev
    Next
    INDEX
    Explanations

    words related to conspiracy and criminal plots

    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
    55
    +0.10
    0.5%
    767
    +0.10
    0.5%
    583
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    767
    +0.10
    0.03
    711
    +0.10
    0.02
    134
    +0.09
    0.03
    Negative Logits
    esa
    -0.83
    eatures
    -0.74
    Flo
    -0.72
    arton
    -0.72
    framework
    -0.71
    ixels
    -0.70
    ixel
    -0.70
    Dialogue
    -0.70
    clamation
    -0.69
    phies
    -0.67
    POSITIVE LOGITS
     sabotage
    1.16
     criminal
    1.11
     murder
    1.07
     conspiring
    1.06
     terror
    1.05
     crime
    1.04
     crimes
    1.03
     extortion
    1.02
     fraud
    1.00
     perpetrated
    1.00
    Activations Density 0.142%

    No Known Activations