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

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Joseph Bloom · Feature Splitting for GPT2-Small
    5. GPT2-Small
    7. Residual Feature Splitting 49152
    9. 8-RES_FS49152-JB
    11. 37556
    Prev
    Next
    INDEX
    Explanations

    discussions around assessment and implementation of policies and procedures

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs49152-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    49,152
    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
    87
    +0.26
    1.3%
    283
    +0.10
    0.5%
    459
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    87
    +0.26
    -0.02
    283
    +0.10
    0.02
    641
    +0.10
    0.05
    Negative Logits
    moon
    -0.69
     Freak
    -0.67
     rejoice
    -0.65
     Horror
    -0.64
    anus
    -0.62
    joy
    -0.62
    famous
    -0.62
     Newsp
    -0.61
    upt
    -0.61
    blogspot
    -0.60
    POSITIVE LOGITS
     impacts
    0.87
     compliance
    0.86
     impact
    0.85
     impacted
    0.85
     appropriate
    0.83
     applicable
    0.81
     applicant
    0.78
     impacting
    0.77
     acceptable
    0.76
     respective
    0.76
    Activations Density 0.456%

    No Known Activations