© 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
    2. Joseph Bloom · Feature Splitting for GPT2-Small
    3. GPT2-Small
    4. Residual Feature Splitting 12288
    5. 8-RES_FS12288-JB
    6. 10267
    Prev
    Next
    INDEX
    Explanations

    information related to technological developments and advancements

    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
    665
    +0.16
    0.8%
    49
    +0.10
    0.5%
    608
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    665
    +0.16
    0.07
    49
    +0.10
    0.06
    608
    +0.10
    0.05
    Negative Logits
    quickShipAvailable
    -0.72
    pite
    -0.69
     Explan
    -0.65
    checking
    -0.64
    hot
    -0.64
    guards
    -0.62
    lies
    -0.60
    checked
    -0.59
    minus
    -0.58
    orno
    -0.58
    POSITIVE LOGITS
     revise
    1.09
     someday
    1.08
     utilize
    1.05
     expand
    1.04
     pursue
    1.04
     bring
    1.02
     continue
    1.02
     introduce
    1.01
     capitalize
    1.00
     improve
    1.00
    Activations Density 0.181%

    No Known Activations