© 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 3072
    9. 8-RES_FS3072-JB
    11. 2990
    Prev
    Next
    INDEX
    Explanations

    quotes or statements made by public figures

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs3072-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    3,072
    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
    725
    +0.12
    0.6%
    481
    +0.10
    0.5%
    470
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    470
    +0.12
    0.07
    430
    +0.10
    0.15
    224
    +0.10
    0.10
    Negative Logits
     ILCS
    -0.76
    brance
    -0.72
     harvest
    -0.71
     sew
    -0.71
     Located
    -0.70
    isite
    -0.69
     LTD
    -0.69
     MAP
    -0.69
    geries
    -0.68
    profits
    -0.66
    POSITIVE LOGITS
     sarcast
    1.31
     rhet
    1.23
     remarks
    1.21
     apologizing
    1.16
     angrily
    1.14
     praising
    1.13
     joking
    1.12
    Asked
    1.10
     apologized
    1.09
     reiterated
    1.08
    Activations Density 3.348%

    No Known Activations