© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    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. 5791
    Prev
    Next
    INDEX
    Explanations

    instances of the word "don't" and its variations

    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
    575
    +0.12
    0.6%
    316
    +0.12
    0.6%
    409
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    696
    +0.12
    0.10
    209
    +0.12
    0.09
    376
    +0.10
    0.07
    Negative Logits
    box
    -0.59
    boxes
    -0.59
    ages
    -0.58
     Veh
    -0.55
    burgh
    -0.55
    wig
    -0.54
    ulously
    -0.54
    guy
    -0.53
     Avalon
    -0.53
     Theodore
    -0.53
    POSITIVE LOGITS
    é¾įå
    0.67
    chieve
    0.66
    icz
    0.63
     cessation
    0.62
    ITION
    0.62
    efe
    0.61
    raught
    0.60
    ENCE
    0.60
     relapse
    0.60
    impact
    0.59
    Activations Density 0.191%

    No Known Activations