Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsBlogSlackPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Joseph Bloom · Feature Splitting for GPT2-Small
    3. GPT2-Small
    4. Residual Feature Splitting 49152
    5. 8-RES_FS49152-JB
    6. 43373
    Prev
    Next
    INDEX
    Explanations

    verbs related to actions, particularly emphasizing constructs of building and the dynamics of doing or taking

    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
    507
    +0.12
    0.5%
    189
    +0.11
    0.5%
    345
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    351
    +0.12
    0.03
    507
    +0.11
    0.04
    189
    +0.10
    0.03
    Negative Logits
    intendent
    -0.76
     awaited
    -0.74
    yrinth
    -0.74
    geoning
    -0.73
    pour
    -0.73
    soon
    -0.72
     finally
    -0.72
    itton
    -0.72
    ortment
    -0.72
    etermined
    -0.69
    POSITIVE LOGITS
     anymore
    1.72
     nor
    1.10
     anything
    1.08
     anywhere
    1.05
     any
    1.02
     anybody
    1.01
     ANY
    1.00
     whatsoever
    0.97
     slightest
    0.96
     necessarily
    0.87
    Activations Density 0.132%

    No Known Activations