© 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
    3. OpenAI · Sparse Autoencoder for GPT2-Small - v5 2024
    5. GPT2-Small
    7. Residual Post - 32k
    9. 8-RES_POST_32K-OAI
    11. 5351
    Prev
    Next
    INDEX
    Explanations

    references to geographical regions, particularly those related to the Mediterranean, Caribbean, and Atlantic

    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
    288
    +0.13
    0.6%
    701
    +0.11
    0.5%
    256
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    288
    +0.13
    0.05
    537
    +0.11
    0.04
    701
    +0.11
    0.03
    Negative Logits
    bol
    -0.90
    perse
    -0.84
    rors
    -0.82
    cler
    -0.81
    ogun
    -0.81
    criptions
    -0.80
    ifice
    -0.80
    anamo
    -0.80
    efer
    -0.80
    etsu
    -0.79
    POSITIVE LOGITS
     Coast
    1.06
     coast
    1.00
     Countries
    0.97
     Regional
    0.96
     Sea
    0.95
     Basin
    0.95
     Territories
    0.94
     region
    0.93
     Peninsula
    0.93
     coasts
    0.93
    Activations Density 0.165%

    No Known Activations