© 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. 19785
    Prev
    Next
    INDEX
    Explanations

    phrases related to speculative statements or opinions

    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
    101
    +0.11
    0.5%
    465
    +0.10
    0.5%
    151
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    465
    +0.11
    0.03
    101
    +0.10
    0.03
    241
    +0.09
    0.03
    Negative Logits
    iak
    -0.79
    efer
    -0.77
    cium
    -0.77
    sheet
    -0.76
    ombat
    -0.75
    bender
    -0.74
    ulative
    -0.74
    iphate
    -0.73
    ulty
    -0.73
    fighters
    -0.71
    POSITIVE LOGITS
     unsurprisingly
    1.13
    haps
    0.83
     sensing
    0.83
     someday
    0.81
     unsur
    0.80
     surprisingly
    0.79
     analogous
    0.76
     indicative
    0.75
     influenced
    0.75
     exacerbated
    0.75
    Activations Density 0.042%

    No Known Activations