© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Joseph Bloom · Feature Splitting for GPT2-Small
    5. GPT2-Small
    7. Residual Feature Splitting 49152
    9. 8-RES_FS49152-JB
    11. 45622
    Prev
    Next
    INDEX
    Explanations

    various numerical data or statistics, particularly related to performance or production metrics

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs49152-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    49,152
    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
    17
    +0.10
    0.4%
    524
    +0.10
    0.4%
    344
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    678
    +0.10
    0.02
    326
    +0.10
    0.02
    392
    +0.09
    0.02
    Negative Logits
    CVE
    -0.65
    reve
    -0.64
     resemb
    -0.63
    abeth
    -0.60
     showc
    -0.60
     scapego
    -0.59
    interpret
    -0.59
    eve
    -0.58
    Spring
    -0.57
    osaurus
    -0.56
    POSITIVE LOGITS
     apiece
    1.25
     increments
    1.11
    +.
    1.03
    +,
    0.98
    illion
    0.97
     per
    0.96
     (~
    0.91
     respectively
    0.90
     (£
    0.89
    ilion
    0.88
    Activations Density 0.049%

    No Known Activations