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. 23921
    Prev
    Next
    INDEX
    Explanations

    phrases that indicate actions or intentions

    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
    608
    +0.12
    0.6%
    508
    +0.12
    0.6%
    90
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    508
    +0.12
    0.05
    718
    +0.12
    0.05
    608
    +0.10
    0.05
    Negative Logits
    living
    -0.78
     fleeing
    -0.72
     dying
    -0.71
     sacrific
    -0.70
    horn
    -0.66
     dies
    -0.64
     sung
    -0.64
    laughter
    -0.63
     hurled
    -0.63
     ringing
    -0.61
    POSITIVE LOGITS
     determine
    1.48
     ensure
    1.43
     optimize
    1.27
     verify
    1.24
     refine
    1.21
     simulate
    1.18
     assure
    1.17
     validate
    1.16
     ascertain
    1.16
     minimize
    1.15
    Activations Density 0.114%

    No Known Activations