© 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. 10-RES_POST_32K-OAI
    11. 32555
    Prev
    Next
    INDEX
    Explanations

    terms that signify implications or indications of certain conditions or facts

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 10-res_post_32k-oai
    Configuration
    jbloom/GPT2-Small-OAI-v5-32k-resid-post-SAEs/v5_32k_layer_10.pt
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    32,768
    Data Type
    torch.float32
    Hook Name
    blocks.10.hook_resid_post
    Hook Layer
    10
    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
    +0.10
    0.5%
    488
    +0.10
    0.5%
    195
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    417
    +0.10
    0.04
    408
    +0.10
    0.04
    236
    +0.10
    0.04
    Negative Logits
    enium
    -0.76
    76561
    -0.75
    uters
    -0.73
    apest
    -0.73
    pton
    -0.72
    »Ĵ
    -0.71
    ighth
    -0.70
    Newsletter
    -0.69
    pez
    -0.69
    quer
    -0.69
    POSITIVE LOGITS
     displeasure
    1.03
     intent
    1.01
     weakness
    0.96
     competence
    0.93
     familiarity
    0.93
     impending
    0.93
     imminent
    0.91
     willingness
    0.90
     disapproval
    0.89
     trouble
    0.88
    Activations Density 0.227%

    No Known Activations