© 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. 764
    Prev
    Next
    INDEX
    Explanations

    names of individuals or locations associated with specific events or stories

    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
    388
    +0.11
    0.5%
    146
    +0.11
    0.5%
    309
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    479
    +0.11
    0.04
    482
    +0.11
    0.04
    284
    +0.10
    0.04
    Negative Logits
    angel
    -0.99
    prus
    -0.96
    aster
    -0.94
    union
    -0.91
    bat
    -0.91
    index
    -0.89
    ija
    -0.89
    space
    -0.88
    silver
    -0.87
    condition
    -0.87
    POSITIVE LOGITS
     M
    1.14
     N
    1.07
     C
    1.04
     H
    1.04
     Y
    1.03
     R
    1.03
     L
    1.03
     S
    1.00
     B
    0.99
     G
    0.99
    Activations Density 0.408%

    No Known Activations