© 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. 20029
    Prev
    Next
    INDEX
    Explanations

    phrases related to personal stories or quotes from individuals

    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
    314
    +0.14
    0.6%
    102
    +0.11
    0.5%
    107
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    675
    +0.14
    0.10
    48
    +0.11
    0.07
    413
    +0.10
    0.07
    Negative Logits
     mushroom
    -0.72
     influential
    -0.70
     incorpor
    -0.66
     wartime
    -0.65
     maximum
    -0.65
     ponder
    -0.64
     targeted
    -0.64
     maximal
    -0.64
     contribution
    -0.63
     assembly
    -0.62
    POSITIVE LOGITS
    cause
    1.12
    ï¸ı
    1.12
    efe
    0.98
    laugh
    0.95
    tis
    0.95
    sung
    0.94
    ¯
    0.93
    nob
    0.91
    she
    0.88
    tu
    0.87
    Activations Density 0.201%

    No Known Activations