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 · Open Source Sparse Autoencoders for all Residual Stream Layers of GPT2-Small
    3. GPT2-Small
    4. Residual Stream
    5. 4-RES-JB
    6. 14982
    Prev
    Next
    INDEX
    Explanations

    instances of blond or blonde hair color

    oai_token-act-pair · gpt-3.5-turbo

    references to people with light-colored hair, particularly blonde individuals

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 4-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.hook_resid_pre
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    24,576
    Data Type
    torch.float32
    Hook Point
    blocks.4.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    4
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    Ö¼
    -0.91
    displayText
    -0.83
    apego
    -0.81
    ablishment
    -0.76
    arters
    -0.76
    ROR
    -0.74
    llah
    -0.74
    GAN
    -0.72
    arnaev
    -0.71
    ADRA
    -0.70
    POSITIVE LOGITS
     wig
    1.23
     blond
    1.15
     blonde
    1.09
     bombshell
    1.08
    haired
    1.05
     hair
    1.04
     hairst
    0.99
    bread
    0.96
     haircut
    0.95
     bob
    0.89
    Activations Density 0.020%

    No Known Activations