© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    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. 5969
    Prev
    Next
    INDEX
    Explanations

    occurrences of the word "master" and its variations, indicative of a focus on mastery or expertise

    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
    207
    +0.09
    0.4%
    373
    +0.09
    0.4%
    136
    +0.08
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    59
    +0.09
    0.02
    288
    +0.09
    0.02
    207
    +0.08
    0.02
    Negative Logits
    enegger
    -0.72
    oldemort
    -0.65
    arf
    -0.64
    heres
    -0.61
    ktop
    -0.59
     DRAG
    -0.58
     toget
    -0.58
    UGE
    -0.58
    AMES
    -0.58
    vernment
    -0.57
    POSITIVE LOGITS
    piece
    1.04
    pieces
    0.85
    Card
    0.82
    Grade
    0.78
    ston
    0.76
    race
    0.75
    mind
    0.75
    planned
    0.74
     sergeant
    0.72
    hood
    0.72
    Activations Density 0.104%

    No Known Activations