© 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. 5719
    Prev
    Next
    INDEX
    Explanations

    occurrences of the word "At" with high activation in various contexts

    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
    314
    +0.18
    0.8%
    365
    +0.13
    0.6%
    28
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    509
    +0.18
    0.03
    59
    +0.13
    0.03
    764
    +0.11
    0.03
    Negative Logits
     Fridays
    -0.61
     Silk
    -0.61
     excise
    -0.60
     Chaser
    -0.60
     Rowling
    -0.57
    theless
    -0.57
     Sundays
    -0.57
     PAY
    -0.56
     Mayweather
    -0.56
     Downing
    -0.55
    POSITIVE LOGITS
    rix
    1.10
    rament
    1.05
    ron
    1.03
    rogen
    1.00
    omy
    0.97
    robe
    0.95
    rod
    0.94
    oka
    0.94
    oms
    0.91
    ro
    0.90
    Activations Density 0.063%

    No Known Activations