© 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 3072
    9. 8-RES_FS3072-JB
    11. 2975
    Prev
    Next
    INDEX
    Explanations

    descriptions of violent incidents or abuse

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs3072-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    3,072
    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
    545
    +0.11
    0.5%
    742
    +0.10
    0.5%
    685
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    96
    +0.11
    0.10
    142
    +0.10
    0.09
    380
    +0.10
    0.04
    Negative Logits
    rastructure
    -0.87
    omics
    -0.85
     sustainability
    -0.84
    osponsors
    -0.84
    erenn
    -0.83
     Innovation
    -0.81
    ahime
    -0.79
    odox
    -0.78
    ocus
    -0.78
     forecasting
    -0.78
    POSITIVE LOGITS
     raping
    1.25
     verbally
    1.16
     raped
    1.16
     sexually
    1.14
     abusive
    1.12
     abuser
    1.12
     bruises
    1.11
     humiliating
    1.10
     orally
    1.10
     humili
    1.10
    Activations Density 3.911%

    No Known Activations