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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    1. Home
    2. OpenAI · Sparse Autoencoder for GPT2-Small - v5 2024
    3. GPT2-Small
    4. Residual Post - 32k
    5. 8-RES_POST_32K-OAI
    6. 4914
    Prev
    Next
    INDEX
    Explanations

    references to individuals' roles and titles in government and politics

    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
    373
    +0.10
    0.4%
    28
    +0.10
    0.4%
    752
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    680
    +0.10
    0.05
    752
    +0.10
    0.09
    362
    +0.09
    0.08
    Negative Logits
     tomorrow
    -0.70
     hereby
    -0.69
    isphere
    -0.68
     hereafter
    -0.67
    bath
    -0.65
     Presence
    -0.62
     learners
    -0.61
    Sov
    -0.59
     heterogeneity
    -0.58
    =-=-=-=-=-=-=-=-
    -0.58
    POSITIVE LOGITS
     unsuccessfully
    1.02
     Nixon
    0.86
     Watergate
    0.80
     intern
    0.77
     Reagan
    0.75
     administrations
    0.75
    oslov
    0.75
     stint
    0.74
    umenthal
    0.73
     briefly
    0.73
    Activations Density 1.455%

    No Known Activations