© 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 12288
    9. 8-RES_FS12288-JB
    11. 746
    Prev
    Next
    INDEX
    Explanations

    words related to immediate actions or responses

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs12288-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    12,288
    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
    565
    +0.10
    0.4%
    144
    +0.10
    0.4%
    755
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    565
    +0.10
    0.03
    105
    +0.10
    0.03
    144
    +0.09
    0.03
    Negative Logits
    smanship
    -0.67
    pel
    -0.66
    glers
    -0.66
    doctor
    -0.65
     ANGEL
    -0.64
    acus
    -0.63
    mic
    -0.62
    CHQ
    -0.60
     Frag
    -0.60
    chance
    -0.60
    POSITIVE LOGITS
    aneously
    1.09
     thereafter
    1.07
    onding
    0.92
     afterward
    0.91
     recognizable
    0.86
     afterwards
    0.85
     regretted
    0.82
     upon
    0.82
     gratification
    0.81
     identifiable
    0.80
    Activations Density 0.049%

    No Known Activations