© 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 49152
    9. 8-RES_FS49152-JB
    11. 41799
    Prev
    Next
    INDEX
    Explanations

    phrases that express levels of familiarity or comfort with a subject

    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
    208
    +0.10
    0.5%
    589
    +0.10
    0.4%
    506
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    117
    +0.10
    0.02
    344
    +0.10
    0.01
    762
    +0.10
    0.02
    Negative Logits
    eworthy
    -0.77
    IRO
    -0.73
     Minor
    -0.70
    eries
    -0.70
    ulo
    -0.69
    yrim
    -0.67
    SpaceEngineers
    -0.66
    thur
    -0.62
    sem
    -0.61
    pmwiki
    -0.61
    POSITIVE LOGITS
    baum
    0.75
     enough
    0.70
    ledge
    0.68
    phrase
    0.67
    enough
    0.65
    aws
    0.65
     summed
    0.64
    é£
    0.63
     forgiven
    0.63
    ledged
    0.63
    Activations Density 0.028%

    No Known Activations