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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & 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. 25766
    Prev
    Next
    INDEX
    Explanations

    references to the concept of dawn or related imagery

    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
    288
    +0.12
    0.5%
    414
    +0.12
    0.5%
    372
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    301
    +0.12
    0.01
    414
    +0.12
    0.01
    372
    +0.10
    0.01
    Negative Logits
    ttle
    -0.79
    chn
    -0.75
    idium
    -0.74
    glomer
    -0.72
    apter
    -0.71
    Wiki
    -0.69
     Mellon
    -0.68
    orie
    -0.67
    akia
    -0.67
    chens
    -0.66
    POSITIVE LOGITS
    ing
    0.95
     dawn
    0.88
    hower
    0.86
    fall
    0.85
    mare
    0.81
    mares
    0.81
    shire
    0.81
     timers
    0.80
    break
    0.77
    ly
    0.77
    Activations Density 0.005%

    No Known Activations