© 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 24576
    9. 8-RES_FS24576-JB
    11. 18935
    Prev
    Next
    INDEX
    Explanations

    plays in a sports context, particularly focused on football

    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
    540
    +0.15
    0.7%
    607
    +0.13
    0.6%
    160
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    540
    +0.15
    0.04
    607
    +0.13
    0.02
    271
    +0.11
    0.02
    Negative Logits
     Beir
    -0.77
    orpor
    -0.66
    izoph
    -0.65
     ink
    -0.64
     whisk
    -0.61
     Unic
    -0.60
    «ĺ
    -0.60
     filament
    -0.59
     Apost
    -0.58
     Reich
    -0.57
    POSITIVE LOGITS
    ername
    1.22
    maker
    1.17
    wright
    1.12
    makers
    1.06
    calling
    1.04
    making
    1.03
    plays
    1.01
    call
    1.00
    writing
    0.96
    offs
    0.95
    Activations Density 0.026%

    No Known Activations