© 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. 1230
    Prev
    Next
    INDEX
    Explanations

    times with hours represented in 'am' or 'pm'

    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
    296
    +0.13
    0.6%
    601
    +0.10
    0.4%
    434
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    296
    +0.13
    0.05
    460
    +0.10
    0.04
    601
    +0.10
    0.03
    Negative Logits
    govern
    -0.72
    chie
    -0.69
     behav
    -0.68
     resemb
    -0.67
    eries
    -0.64
    gotten
    -0.63
     acknowled
    -0.63
    proof
    -0.63
    designed
    -0.59
    bons
    -0.58
    POSITIVE LOGITS
     EDT
    1.40
     EST
    1.37
     PDT
    1.28
     PST
    1.27
     CET
    1.21
     CST
    1.18
     ET
    1.18
     GMT
    1.15
     PT
    1.03
     UTC
    1.02
    Activations Density 0.082%

    No Known Activations