© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW Community BlogPrivacy & TermsContact
    1. Home
    3. OpenAI · Sparse Autoencoder for GPT2-Small - v5 2024
    5. GPT2-Small
    7. Residual Post - 32k
    9. 10-RES_POST_32K-OAI
    11. 26244
    Prev
    Next
    INDEX
    Explanations

    terms related to aggregation and collective attributes

    oai_token-act-pair · gpt-4o-mini
    Triggered 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
    Neuron Alignment
    Index
    Value
    % of L₁
    710
    +0.14
    0.6%
    206
    +0.13
    0.6%
    590
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    591
    +0.14
    0.01
    107
    +0.13
    0.01
    103
    +0.10
    0.01
    Negative Logits
    erence
    -0.80
    endi
    -0.78
    ffen
    -0.72
    byter
    -0.71
    erity
    -0.71
    ensional
    -0.70
    ence
    -0.69
    pta
    -0.69
    pter
    -0.68
    ppe
    -0.67
    POSITIVE LOGITS
    amine
    0.74
     guiName
    0.74
    ãĥķãĤ¡
    0.72
    omal
    0.71
    ãĥŃ
    0.70
    ãĥĸ
    0.69
    olor
    0.68
    ãĤ¤ãĥĪ
    0.67
    ATA
    0.67
    ãĥĪ
    0.67
    Activations Density 0.053%

    No Known Activations