© 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. 41723
    Prev
    Next
    INDEX
    Explanations

    the word "as" in various contexts

    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
    235
    +0.09
    0.4%
    445
    +0.09
    0.4%
    393
    +0.08
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    647
    +0.09
    0.05
    172
    +0.09
    0.05
    192
    +0.08
    0.04
    Negative Logits
    itiveness
    -0.70
    ople
    -0.66
    OY
    -0.65
    atche
    -0.60
    sonian
    -0.60
     Mellon
    -0.59
    OWS
    -0.58
    ivas
    -0.57
     resurgence
    -0.57
    åħ
    -0.57
    POSITIVE LOGITS
     opposed
    1.34
     well
    1.09
    ynchron
    1.06
     evidenced
    1.03
     shown
    1.02
    bestos
    1.02
     follows
    1.02
    cription
    1.00
    pects
    0.95
    well
    0.95
    Activations Density 0.099%

    No Known Activations