© 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
    2. OpenAI · Sparse Autoencoder for GPT2-Small - v5 2024
    3. GPT2-Small
    4. Residual Post - 32k
    5. 8-RES_POST_32K-OAI
    6. 4564
    Prev
    Next
    INDEX
    Explanations

    words that indicate connections or relationships between concepts or entities

    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
    559
    +0.12
    0.5%
    29
    +0.10
    0.5%
    187
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    611
    +0.12
    0.04
    665
    +0.10
    0.04
    1
    +0.10
    0.03
    Negative Logits
    bis
    -0.76
    lvl
    -0.75
    achment
    -0.68
    ety
    -0.68
    cheat
    -0.68
    dar
    -0.65
    estate
    -0.64
    renheit
    -0.62
    sv
    -0.62
    lied
    -0.61
    POSITIVE LOGITS
     dots
    1.18
     disparate
    1.03
     sexes
    0.80
     them
    0.69
     oneself
    0.69
     together
    0.68
     apples
    0.67
     seamlessly
    0.64
     genders
    0.61
    ombs
    0.61
    Activations Density 0.361%

    No Known Activations