© 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 12288
    9. 8-RES_FS12288-JB
    11. 9709
    Prev
    Next
    INDEX
    Explanations

    user interactions and data analysis in a technical or organizational context

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs12288-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    12,288
    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
    447
    +0.17
    0.8%
    33
    +0.10
    0.5%
    495
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    288
    +0.17
    0.07
    51
    +0.10
    0.06
    575
    +0.10
    0.06
    Negative Logits
    cffffcc
    -0.93
    ãĤ´ãĥ³
    -0.81
    çīĪ
    -0.66
    eln
    -0.65
    eatures
    -0.59
    ĪĴ
    -0.59
    ebin
    -0.58
     guise
    -0.57
    awks
    -0.56
    ursday
    -0.56
    POSITIVE LOGITS
     available
    0.85
     involved
    0.83
     per
    0.80
     consumed
    0.77
     allocated
    0.76
     required
    0.76
     devoted
    0.75
     varies
    0.75
     surveyed
    0.72
     needed
    0.72
    Activations Density 0.444%

    No Known Activations