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

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & 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. 46009
    Prev
    Next
    INDEX
    Explanations

    phrases related to relational dynamics and personal inquiries

    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
    218
    +0.12
    0.5%
    363
    +0.12
    0.5%
    481
    +0.12
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    227
    +0.12
    0.02
    363
    +0.12
    0.03
    351
    +0.12
    0.02
    Negative Logits
    isively
    -0.81
    ilib
    -0.78
    ords
    -0.74
    onomy
    -0.74
    atories
    -0.72
     Canaver
    -0.69
    rates
    -0.66
    ocr
    -0.65
    acket
    -0.65
    estyles
    -0.65
    POSITIVE LOGITS
     spouse
    0.99
     unexpected
    0.97
     overheard
    0.97
    Someone
    0.96
    someone
    0.95
     delinquent
    0.91
     babys
    0.90
     misplaced
    0.89
     gossip
    0.89
     misunderstood
    0.89
    Activations Density 0.282%

    No Known Activations