© 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. 19997
    Prev
    Next
    INDEX
    Explanations

    phrases related to baseball statistics - specifically RBIs and home runs

    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
    254
    +0.11
    0.5%
    504
    +0.10
    0.4%
    505
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    505
    +0.11
    0.02
    500
    +0.10
    0.02
    326
    +0.10
    0.01
    Negative Logits
    urnal
    -0.79
    ĨĴ
    -0.71
    MSN
    -0.69
     srfAttach
    -0.68
     Shant
    -0.67
    enegger
    -0.67
     consultation
    -0.66
    ffect
    -0.64
     ethn
    -0.63
    æ©Ł
    -0.63
    POSITIVE LOGITS
     scored
    1.06
     batted
    1.04
     scorer
    0.90
     hitters
    0.90
     hitter
    0.88
     scoring
    0.83
    gers
    0.81
     steal
    0.79
     RBI
    0.79
     batters
    0.79
    Activations Density 0.026%

    No Known Activations