Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsBlogSlackPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Joseph Bloom · Feature Splitting for GPT2-Small
    3. GPT2-Small
    4. Residual Feature Splitting 768
    5. 8-RES_FS768-JB
    6. 29
    Prev
    Next
    INDEX
    Explanations

    comparisons or contrasts

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs768-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    768
    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
    373
    +0.12
    0.5%
    223
    +0.10
    0.4%
    84
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    223
    +0.12
    0.09
    327
    +0.10
    0.07
    662
    +0.09
    0.07
    Negative Logits
    renheit
    -0.75
    nell
    -0.70
    ilit
    -0.69
    xus
    -0.69
    biz
    -0.69
    vous
    -0.69
    plete
    -0.68
    vich
    -0.67
    scrib
    -0.66
    uably
    -0.66
    POSITIVE LOGITS
     sexes
    1.68
     sides
    1.44
     halves
    1.40
     genders
    1.37
     parties
    0.96
     extremes
    0.92
     coasts
    0.89
     Houses
    0.88
     ends
    0.86
     thirds
    0.82
    Activations Density 2.436%

    No Known Activations