© 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. 28994
    Prev
    Next
    INDEX
    Explanations

    instances of newsworthy events or stories that gain significant public attention

    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
    378
    +0.14
    0.6%
    94
    +0.12
    0.5%
    174
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    547
    +0.14
    0.06
    426
    +0.12
    0.04
    245
    +0.11
    0.10
    Negative Logits
     maintenance
    -0.68
     upkeep
    -0.65
    atto
    -0.65
    ³³³³³³³³³³³³³³³³
    -0.62
     stationed
    -0.62
     Cth
    -0.62
    Condition
    -0.61
    :]
    -0.61
     Intro
    -0.60
     maintaining
    -0.59
    POSITIVE LOGITS
     headlines
    1.29
     news
    0.94
     worldwide
    0.86
     herald
    0.85
     tabl
    0.84
    waves
    0.84
     headline
    0.83
     sensational
    0.83
     tabloid
    0.83
    CNN
    0.83
    Activations Density 4.502%

    No Known Activations