© 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. 19906
    Prev
    Next
    INDEX
    Explanations

    phrases related to journalism and news reporting

    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
    616
    +0.13
    0.6%
    362
    +0.11
    0.5%
    393
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    362
    +0.13
    0.04
    473
    +0.11
    0.03
    616
    +0.11
    0.03
    Negative Logits
    abi
    -0.78
    ibaba
    -0.75
    amsung
    -0.70
     propagate
    -0.68
    pport
    -0.65
    etsk
    -0.63
    agascar
    -0.63
    abee
    -0.61
    ochond
    -0.60
    iatrics
    -0.59
    POSITIVE LOGITS
     end
    0.92
     margins
    0.85
     sheer
    0.84
     Numbers
    0.84
     way
    0.77
     implication
    0.76
     standards
    0.76
     virtue
    0.75
     same
    0.73
     fireplace
    0.71
    Activations Density 0.065%

    No Known Activations