© 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. 7052
    Prev
    Next
    INDEX
    Explanations

    phrases that indicate the act of revealing or uncovering information

    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
    101
    +0.11
    0.5%
    642
    +0.10
    0.5%
    610
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    101
    +0.11
    0.03
    642
    +0.10
    0.02
    195
    +0.09
    0.01
    Negative Logits
    yip
    -0.74
     captcha
    -0.72
     paced
    -0.62
    oslav
    -0.62
    creation
    -0.60
    asus
    -0.60
    create
    -0.59
    otor
    -0.59
    anwhile
    -0.58
    isol
    -0.58
    POSITIVE LOGITS
     loopholes
    0.93
     secrets
    0.87
     truths
    0.87
    ibility
    0.85
    orial
    0.82
    iary
    0.75
    ively
    0.73
    arian
    0.72
    heny
    0.70
    ered
    0.70
    Activations Density 0.061%

    No Known Activations