© 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. 6537
    Prev
    Next
    INDEX
    Explanations

    proper nouns related to sports figures, particularly coaches and managers

    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
    170
    +0.13
    0.6%
    415
    +0.10
    0.5%
    753
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    659
    +0.13
    0.10
    224
    +0.10
    0.07
    502
    +0.09
    0.06
    Negative Logits
     actress
    -0.72
     protester
    -0.72
     Sakuya
    -0.72
    acters
    -0.72
     prostitute
    -0.71
     Intercept
    -0.70
     Sabha
    -0.70
    ertodd
    -0.69
     Spectre
    -0.69
     Crescent
    -0.69
    POSITIVE LOGITS
     coaching
    1.01
     coached
    0.99
     oversaw
    0.98
    ichick
    0.93
     instituted
    0.90
     overseeing
    0.89
    management
    0.86
     administ
    0.86
     coaches
    0.85
    swer
    0.85
    Activations Density 1.046%

    No Known Activations