© 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. 26323
    Prev
    Next
    INDEX
    Explanations

    instructions related to digital file management and processes

    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
    373
    +0.18
    0.9%
    712
    +0.12
    0.5%
    224
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    326
    +0.18
    0.08
    297
    +0.12
    0.04
    519
    +0.10
    0.03
    Negative Logits
     disbanded
    -0.77
     dispers
    -0.65
    ysis
    -0.64
     retracted
    -0.63
    NPR
    -0.61
     Decay
    -0.61
     fading
    -0.60
     Mour
    -0.59
     starving
    -0.58
    ilee
    -0.57
    POSITIVE LOGITS
     Scroll
    0.81
    Scroll
    0.77
    microsoft
    0.77
    Under
    0.75
     Under
    0.73
    button
    0.72
    og
    0.72
    select
    0.71
     scroll
    0.71
    look
    0.71
    Activations Density 0.974%

    No Known Activations