© 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. 26572
    Prev
    Next
    INDEX
    Explanations

    phrases that indicate feasibility or likelihood

    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
    151
    +0.13
    0.6%
    221
    +0.11
    0.5%
    348
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    221
    +0.13
    0.01
    151
    +0.11
    0.01
    348
    +0.10
    0.01
    Negative Logits
    uay
    -0.79
    abase
    -0.75
    lua
    -0.74
    sbm
    -0.72
    dro
    -0.70
    gar
    -0.70
    rets
    -0.68
    alsa
    -0.68
    alg
    -0.67
    masters
    -0.66
    POSITIVE LOGITS
     intervene
    0.75
    iary
    0.74
     accompanied
    0.74
     utilize
    0.70
     avoid
    0.68
     capitalize
    0.68
     recourse
    0.68
     practicable
    0.67
     relocate
    0.65
     opt
    0.65
    Activations Density 0.012%

    No Known Activations