© 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 49152
    9. 8-RES_FS49152-JB
    11. 41677
    Prev
    Next
    INDEX
    Explanations

    references to official briefings or updates

    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
    679
    +0.17
    0.8%
    288
    +0.16
    0.7%
    195
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    679
    +0.17
    0.01
    115
    +0.16
    0.01
    729
    +0.09
    0.01
    Negative Logits
     Abram
    -0.67
    ymes
    -0.66
    otrop
    -0.65
     adultery
    -0.64
    amaz
    -0.64
    relative
    -0.63
    alde
    -0.63
    aden
    -0.61
    negie
    -0.61
    acea
    -0.60
    POSITIVE LOGITS
     briefings
    1.17
     briefing
    1.05
     sessions
    0.91
     convened
    0.84
     session
    0.82
     room
    0.78
     attendees
    0.78
     Papers
    0.77
     slides
    0.77
     booth
    0.75
    Activations Density 0.005%

    No Known Activations