© 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 EvalsExportsinterp-engineNEWAPI 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. 7257
    Prev
    Next
    INDEX
    Explanations

    negative connotations or downsides associated with various subjects

    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
    707
    +0.10
    0.5%
    179
    +0.10
    0.4%
    680
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    314
    +0.10
    0.01
    152
    +0.10
    0.04
    83
    +0.10
    0.03
    Negative Logits
     barrier
    -0.79
     gamb
    -0.74
     swirl
    -0.73
     bump
    -0.72
     inhabited
    -0.71
     deval
    -0.70
     hiding
    -0.69
     hole
    -0.69
     novelist
    -0.69
     rav
    -0.68
    POSITIVE LOGITS
    TV
    1.03
    FM
    0.96
    Reloaded
    0.95
    HQ
    0.94
    KA
    0.93
    USA
    0.91
    agents
    0.90
    2018
    0.90
    Radio
    0.89
    EVA
    0.88
    Activations Density 0.186%

    No Known Activations