© 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. 41715
    Prev
    Next
    INDEX
    Explanations

    words related to the concept of holding or storage capacity

    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
    206
    +0.10
    0.5%
    183
    +0.10
    0.4%
    254
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    206
    +0.10
    0.02
    582
    +0.10
    0.02
    531
    +0.10
    0.02
    Negative Logits
    hedon
    -0.73
    DIT
    -0.73
    clerosis
    -0.72
    ffe
    -0.72
    shire
    -0.71
    vious
    -0.69
    isode
    -0.68
    DragonMagazine
    -0.68
    schild
    -0.68
    ixel
    -0.67
    POSITIVE LOGITS
    hold
    1.01
     sway
    1.01
    holders
    0.98
     hold
    0.93
     hostage
    0.91
    holding
    0.88
    holder
    0.86
    erness
    0.84
     onto
    0.83
     steady
    0.83
    Activations Density 0.021%

    No Known Activations