© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI 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. 18462
    Prev
    Next
    INDEX
    Explanations

    the verb 'help'

    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
    373
    +0.15
    0.7%
    718
    +0.11
    0.5%
    218
    +0.09
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    112
    +0.15
    0.01
    718
    +0.11
    0.01
    750
    +0.09
    0.01
    Negative Logits
    theless
    -0.64
    etheus
    -0.63
    andom
    -0.62
    rane
    -0.60
    sonian
    -0.57
    avior
    -0.56
     threat
    -0.54
     unknown
    -0.54
    bon
    -0.51
     bounty
    -0.51
    POSITIVE LOGITS
     noticing
    0.66
    ctor
    0.65
    ãĤ¨ãĥ«
    0.58
     answering
    0.57
    ãĤ®
    0.57
    peak
    0.56
     fielding
    0.55
     imagining
    0.55
     Sega
    0.55
     anybody
    0.54
    Activations Density 0.023%

    No Known Activations