© 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
    3. OpenAI · Sparse Autoencoder for GPT2-Small - v5 2024
    5. GPT2-Small
    7. Residual Post - 32k
    9. 10-RES_POST_32K-OAI
    11. 28130
    Prev
    Next
    INDEX
    Explanations

    instances of correspondence or communication, particularly written letters and emails

    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
    547
    +0.13
    0.6%
    347
    +0.11
    0.5%
    275
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    608
    +0.13
    0.07
    59
    +0.11
    0.04
    121
    +0.10
    0.05
    Negative Logits
    igmatic
    -0.61
    gal
    -0.61
     Neph
    -0.60
    rifice
    -0.60
     Ended
    -0.60
     Magicka
    -0.59
    bender
    -0.59
    avascript
    -0.59
    natureconservancy
    -0.58
    arson
    -0.58
    POSITIVE LOGITS
     editor
    0.97
     Editor
    0.96
     editors
    0.95
     inbox
    0.92
     informing
    0.85
     superiors
    0.84
     subscribers
    0.81
     urging
    0.79
     thanking
    0.78
     asking
    0.78
    Activations Density 0.602%

    No Known Activations