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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    1. Home
    2. OpenAI · Sparse Autoencoder for GPT2-Small - v5 2024
    3. GPT2-Small
    4. Residual Post - 32k
    5. 8-RES_POST_32K-OAI
    6. 5256
    Prev
    Next
    INDEX
    Explanations

    content related to security vulnerabilities and attacks

    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
    536
    +0.10
    0.5%
    655
    +0.09
    0.4%
    39
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    718
    +0.10
    0.13
    49
    +0.09
    0.12
    665
    +0.09
    0.11
    Negative Logits
    ugen
    -0.73
    lite
    -0.71
    ãĥĺãĥ©
    -0.69
    tek
    -0.69
    escent
    -0.68
    STEM
    -0.68
     Cities
    -0.67
    engineering
    -0.67
    horn
    -0.67
    ryu
    -0.66
    POSITIVE LOGITS
     eaves
    1.05
     instantly
    0.95
     retri
    0.93
     spam
    0.92
     debugging
    0.90
     pinpoint
    0.90
     inaccurate
    0.88
     retrieval
    0.88
     cached
    0.87
     securely
    0.87
    Activations Density 3.176%

    No Known Activations