© 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. 348
    Prev
    Next
    INDEX
    Explanations

    pairs of opposite terms

    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
    314
    +0.10
    0.5%
    284
    +0.10
    0.5%
    305
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    284
    +0.10
    0.07
    393
    +0.10
    0.06
    384
    +0.10
    0.04
    Negative Logits
    igraph
    -0.70
     veins
    -0.68
    ipel
    -0.64
    éĹĺ
    -0.64
    oker
    -0.64
    mun
    -0.63
    ETHOD
    -0.61
     Registry
    -0.60
    bay
    -0.59
    erey
    -0.58
    POSITIVE LOGITS
     downside
    0.78
    disable
    0.67
     subtract
    0.66
     negatives
    0.65
     indifferent
    0.65
     sqor
    0.64
    value
    0.63
    Ô
    0.63
     pessim
    0.62
     FALSE
    0.62
    Activations Density 0.497%

    No Known Activations