© 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. 18783
    Prev
    Next
    INDEX
    Explanations

    phrases related to acting in accordance with the correct or optimal way

    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
    278
    +0.10
    0.5%
    56
    +0.09
    0.4%
    373
    +0.08
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    278
    +0.10
    0.03
    81
    +0.09
    0.03
    21
    +0.08
    0.03
    Negative Logits
    ĸļ
    -0.82
    anned
    -0.77
    BuyableInstoreAndOnline
    -0.68
    bery
    -0.67
    cit
    -0.63
    avia
    -0.63
     Rica
    -0.63
    ushima
    -0.62
    ADRA
    -0.62
     conclud
    -0.60
    POSITIVE LOGITS
    eous
    0.87
     amount
    0.86
     wing
    0.86
     combination
    0.82
     thing
    0.81
     kind
    0.80
     circumstances
    0.80
     way
    0.80
     ballpark
    0.80
     attitude
    0.76
    Activations Density 0.034%

    No Known Activations