© 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. 30684
    Prev
    Next
    INDEX
    Explanations

    phrases and terms related to athletic performance and evaluation

    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
    540
    +0.14
    0.7%
    300
    +0.12
    0.6%
    426
    +0.12
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    540
    +0.14
    0.08
    271
    +0.12
    0.06
    188
    +0.12
    0.05
    Negative Logits
     spoilers
    -0.78
    reddits
    -0.74
     Laksh
    -0.72
    yll
    -0.70
    leanor
    -0.70
     Ladies
    -0.68
     Pigs
    -0.68
     Lands
    -0.68
    ezvous
    -0.67
     Worse
    -0.65
    POSITIVE LOGITS
     athleticism
    1.24
     explos
    1.05
     instincts
    1.01
     agility
    1.00
     upside
    0.95
     toughness
    0.95
     speed
    0.93
     maturity
    0.93
     athletic
    0.92
     versatility
    0.91
    Activations Density 0.890%

    No Known Activations