© 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. 26548
    Prev
    Next
    INDEX
    Explanations

    statistical or numerical data related to performance metrics

    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
    62
    +0.14
    0.6%
    404
    +0.09
    0.4%
    208
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    538
    +0.14
    0.01
    326
    +0.09
    0.01
    628
    +0.09
    0.01
    Negative Logits
    ocene
    -0.85
     atmosp
    -0.66
     advis
    -0.66
     reconc
    -0.65
     cir
    -0.65
     caut
    -0.63
    ously
    -0.63
     Nights
    -0.63
     toget
    -0.62
     nerv
    -0.62
    POSITIVE LOGITS
    anish
    0.91
    ARGET
    0.91
    roit
    0.89
    uple
    0.88
    LA
    0.85
    NI
    0.85
    ION
    0.84
    uning
    0.83
    rans
    0.83
    ECH
    0.82
    Activations Density 0.005%

    No Known Activations