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

    Neuronpedia

    Jacobian LensNEW
    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. 6855
    Prev
    Next
    INDEX
    Explanations

    actions and processes related to creation, modification, and configuration

    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
    103
    +0.13
    0.6%
    184
    +0.10
    0.4%
    306
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    665
    +0.13
    0.12
    435
    +0.10
    0.06
    678
    +0.09
    0.09
    Negative Logits
    ãĤ©
    -0.62
    Defense
    -0.61
    BuyableInstoreAndOnline
    -0.59
    lights
    -0.58
     ãĤµãĥ¼ãĥĨãĤ£ãĥ¯ãĥ³
    -0.58
    \":
    -0.57
    ĸļ
    -0.57
    Ry
    -0.56
    los
    -0.55
     coincided
    -0.55
    POSITIVE LOGITS
     anytime
    1.53
     anywhere
    1.32
     easily
    1.00
     anonymously
    0.96
     indefinitely
    0.95
     freely
    0.92
     any
    0.91
     anything
    0.88
     whatever
    0.87
     unlimited
    0.87
    Activations Density 4.336%

    No Known Activations