© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. OpenAI · Sparse Autoencoder for GPT2-Small - v5 2024
    5. GPT2-Small
    7. Residual Post - 32k
    9. 8-RES_POST_32K-OAI
    11. 6099
    Prev
    Next
    INDEX
    Explanations

    terms and phrases related to fantasy or supernatural elements

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_post_32k-oai
    Configuration
    jbloom/GPT2-Small-OAI-v5-32k-resid-post-SAEs/v5_32k_layer_8.pt
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    32,768
    Data Type
    torch.float32
    Hook Name
    blocks.8.hook_resid_post
    Hook Layer
    8
    Architecture
    standard
    Context Size
    64
    Dataset
    Skylion007/openwebtext
    Activation Function
    topk
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    640
    +0.11
    0.5%
    588
    +0.11
    0.5%
    176
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    640
    +0.11
    0.12
    6
    +0.11
    0.11
    176
    +0.10
    0.06
    Negative Logits
    ographically
    -0.75
    actly
    -0.72
     translating
    -0.72
    cember
    -0.71
    desc
    -0.70
     shares
    -0.70
     materially
    -0.69
    ometime
    -0.67
     gains
    -0.67
     abundantly
    -0.66
    POSITIVE LOGITS
     Balloon
    0.91
     Compass
    0.89
     Smooth
    0.89
     Shark
    0.89
     Electro
    0.87
     Laser
    0.86
     Desk
    0.86
     Anch
    0.86
     Gy
    0.86
     Rhino
    0.85
    Activations Density 0.742%

    No Known Activations