Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Joseph Bloom · Feature Splitting for GPT2-Small
    3. GPT2-Small
    4. Residual Feature Splitting 49152
    5. 8-RES_FS49152-JB
    6. 25874
    Prev
    Next
    INDEX
    Explanations

    references to robbery or theft-related concepts

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs49152-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    49,152
    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
    707
    +0.10
    0.5%
    174
    +0.10
    0.5%
    316
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    174
    +0.10
    0.01
    376
    +0.10
    0.01
    521
    +0.10
    0.01
    Negative Logits
     largeDownload
    -0.76
    alam
    -0.71
    hips
    -0.67
     point
    -0.64
    cape
    -0.63
     depiction
    -0.62
     redes
    -0.60
    jad
    -0.58
     è£ıè¦ļéĨĴ
    -0.57
    chwitz
    -0.57
    POSITIVE LOGITS
    bing
    1.15
    acter
    1.07
    atically
    1.06
    atic
    0.96
    rities
    0.95
    inson
    0.94
    ility
    0.94
    ooth
    0.92
    abies
    0.91
    bers
    0.89
    Activations Density 0.004%

    No Known Activations