© 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. 37339
    Prev
    Next
    INDEX
    Explanations

    references to fouls in sports contexts

    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
    63
    +0.13
    0.6%
    745
    +0.12
    0.5%
    167
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    63
    +0.13
    0.01
    47
    +0.12
    0.01
    487
    +0.10
    0.01
    Negative Logits
    ocobo
    -0.72
    akeru
    -0.72
    Downloadha
    -0.72
    edia
    -0.71
    _>
    -0.69
    udeau
    -0.64
     Airl
    -0.64
    itan
    -0.64
    btn
    -0.63
    1920
    -0.63
    POSITIVE LOGITS
    cery
    0.96
    terness
    0.88
    sie
    0.84
     smelling
    0.79
    s
    0.79
    nesses
    0.79
    mouth
    0.77
    ness
    0.76
    eners
    0.72
    ster
    0.71
    Activations Density 0.005%

    No Known Activations