© 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 EvalsExportsinterp-engineNEWAPI 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. 7860
    Prev
    Next
    INDEX
    Explanations

    possessive pronouns and references to personal relationships or emotions

    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
    517
    +0.13
    0.6%
    532
    +0.12
    0.5%
    258
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    362
    +0.13
    0.08
    92
    +0.12
    0.05
    503
    +0.09
    0.04
    Negative Logits
     Canaver
    -0.72
     provoking
    -0.65
     dup
    -0.65
    ilitating
    -0.60
    reath
    -0.60
     neighboring
    -0.58
     Boost
    -0.57
    agos
    -0.57
    amazon
    -0.56
    lette
    -0.55
    POSITIVE LOGITS
     intention
    0.93
     opinion
    0.89
     intentions
    0.86
     opin
    0.84
     intent
    0.83
     displeasure
    0.81
     innocence
    0.80
     condolences
    0.79
     frustrations
    0.79
     disapproval
    0.78
    Activations Density 0.340%

    No Known Activations