© 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. 5073
    Prev
    Next
    INDEX
    Explanations

    statements about scientific validity and the interpretation of research findings

    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
    705
    +0.11
    0.5%
    327
    +0.10
    0.5%
    80
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    718
    +0.11
    0.16
    540
    +0.10
    0.15
    208
    +0.10
    0.12
    Negative Logits
    illet
    -0.69
     gays
    -0.67
     mirac
    -0.66
     Duchess
    -0.65
     Purg
    -0.64
     homosexuals
    -0.63
     cure
    -0.63
    washed
    -0.62
     cav
    -0.62
     magically
    -0.62
    POSITIVE LOGITS
     anecdotal
    1.07
     unlikely
    1.03
     anecd
    1.00
     situational
    0.99
     certainly
    0.93
     speculative
    0.92
     typically
    0.87
     observational
    0.86
     likely
    0.85
     uncertainties
    0.84
    Activations Density 3.684%

    No Known Activations