© 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
    2. OpenAI · Sparse Autoencoder for GPT2-Small - v5 2024
    3. GPT2-Small
    4. Residual Post - 32k
    5. 10-RES_POST_32K-OAI
    6. 32537
    Prev
    Next
    INDEX
    Explanations

    numerical statistics and performance metrics

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 10-res_post_32k-oai
    Configuration
    jbloom/GPT2-Small-OAI-v5-32k-resid-post-SAEs/v5_32k_layer_10.pt
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    32,768
    Data Type
    torch.float32
    Hook Name
    blocks.10.hook_resid_post
    Hook Layer
    10
    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
    211
    +0.11
    0.5%
    388
    +0.10
    0.4%
    27
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    539
    +0.11
    0.18
    442
    +0.10
    0.17
    640
    +0.09
    0.15
    Negative Logits
    ammers
    -0.66
     SX
    -0.63
    oln
    -0.62
     CTR
    -0.62
    DAQ
    -0.61
     Barclays
    -0.59
    rador
    -0.59
     CASE
    -0.58
     LU
    -0.57
     FOIA
    -0.57
    POSITIVE LOGITS
    ãĤ¼
    0.71
     forgetting
    0.66
    sav
    0.66
    isine
    0.64
    !:
    0.62
    alty
    0.61
     aver
    0.61
     Goodbye
    0.60
     Translation
    0.60
    etime
    0.60
    Activations Density 1.596%

    No Known Activations