© 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. 10-RES_POST_32K-OAI
    11. 28685
    Prev
    Next
    INDEX
    Explanations

    hyphenated expressions and numerical data in sports or other competitive contexts

    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
    102
    +0.14
    0.6%
    176
    +0.13
    0.6%
    360
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    102
    +0.14
    0.02
    560
    +0.13
    0.01
    160
    +0.11
    0.02
    Negative Logits
     hourly
    -0.78
     elementary
    -0.68
     average
    -0.66
     masculinity
    -0.65
     hour
    -0.65
     etiquette
    -0.64
     itch
    -0.64
     mediation
    -0.63
     Corpus
    -0.61
     royalty
    -0.61
    POSITIVE LOGITS
    trade
    1.14
    clear
    1.13
    brid
    1.11
    position
    1.07
    unders
    1.07
    shoot
    1.06
    items
    1.05
    display
    1.04
    miss
    1.03
    done
    1.03
    Activations Density 0.103%

    No Known Activations