© 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 12288
    9. 8-RES_FS12288-JB
    11. 856
    Prev
    Next
    INDEX
    Explanations

    phrases related to events or procedures involving a first-come, first-served system

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs12288-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    12,288
    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
    717
    +0.12
    0.5%
    290
    +0.11
    0.5%
    394
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    290
    +0.12
    0.04
    717
    +0.11
    0.03
    460
    +0.10
    0.03
    Negative Logits
    abled
    -0.79
    arine
    -0.77
    FORMATION
    -0.72
    ingen
    -0.72
    intosh
    -0.71
    sung
    -0.67
    interrupted
    -0.67
    xual
    -0.66
    chrom
    -0.66
    rw
    -0.66
    POSITIVE LOGITS
     foremost
    0.99
    imester
    0.80
    ©¶æ
    0.76
     Misty
    0.74
     introdu
    0.66
     princ
    0.65
     lesson
    0.64
    isite
    0.63
     Nicola
    0.62
    ttle
    0.62
    Activations Density 0.180%

    No Known Activations