© 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. 19871
    Prev
    Next
    INDEX
    Explanations

    the word "why" used in a context where the speaker expresses confusion or lack of understanding

    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
    750
    +0.12
    0.6%
    533
    +0.10
    0.5%
    679
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    750
    +0.12
    0.03
    533
    +0.10
    0.03
    628
    +0.10
    0.02
    Negative Logits
    lator
    -0.85
    aughed
    -0.77
    ymph
    -0.68
    haul
    -0.65
    iece
    -0.65
    opic
    -0.65
    ibaba
    -0.64
    thal
    -0.63
    Runner
    -0.63
     Roller
    -0.62
    POSITIVE LOGITS
     bother
    0.83
    soever
    0.82
     someone
    0.79
     somebody
    0.78
     why
    0.75
     exactly
    0.72
     they
    0.71
     anyone
    0.71
     people
    0.70
     anybody
    0.68
    Activations Density 0.038%

    No Known Activations