© 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 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. 6207
    Prev
    Next
    INDEX
    Explanations

    distinctions between different categories of entities or concepts, particularly focusing on non-entities versus their contrasted counterparts

    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
    464
    +0.14
    0.6%
    314
    +0.12
    0.6%
    760
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    102
    +0.14
    0.07
    314
    +0.12
    0.01
    616
    +0.09
    0.04
    Negative Logits
     Panic
    -0.72
     Aval
    -0.72
     Carnage
    -0.68
     Tosh
    -0.68
     Cthulhu
    -0.65
     Saturn
    -0.65
     Comet
    -0.64
     flourish
    -0.64
     Swords
    -0.64
     Twist
    -0.63
    POSITIVE LOGITS
     versa
    0.92
    jad
    0.82
    aque
    0.80
    existent
    0.78
    acqu
    0.76
     thereof
    0.76
    org
    0.76
    idden
    0.75
    chron
    0.75
    juven
    0.75
    Activations Density 0.230%

    No Known Activations