© 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. 8-RES_POST_32K-OAI
    11. 6146
    Prev
    Next
    INDEX
    Explanations

    dialogue or conversational elements where questions are asked or responses are given

    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
    563
    +0.11
    0.5%
    278
    +0.10
    0.5%
    168
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    6
    +0.11
    0.09
    472
    +0.10
    0.03
    233
    +0.09
    0.05
    Negative Logits
    İĭ
    -0.73
    arde
    -0.73
    ronics
    -0.72
     coping
    -0.65
     rhy
    -0.64
     sket
    -0.62
    ource
    -0.58
    iper
    -0.57
     incompetence
    -0.57
     insertion
    -0.57
    POSITIVE LOGITS
     yes
    1.80
    yes
    1.76
    Yes
    1.69
     YES
    1.67
     Yes
    1.65
    YES
    1.53
    Absolutely
    1.50
     Nope
    1.38
     Absolutely
    1.33
    Yep
    1.24
    Activations Density 1.187%

    No Known Activations