© 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. 10-RES_POST_32K-OAI
    11. 30481
    Prev
    Next
    INDEX
    Explanations

    the word "only" and its variations

    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
    628
    +0.12
    0.6%
    756
    +0.12
    0.5%
    155
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    628
    +0.12
    0.06
    498
    +0.12
    0.06
    524
    +0.11
    0.05
    Negative Logits
    hement
    -0.70
     Conclusion
    -0.69
    idon
    -0.63
    phen
    -0.61
    zin
    -0.60
    ikuman
    -0.59
     Monarch
    -0.58
     Everest
    -0.57
    bis
    -0.56
    gen
    -0.56
    POSITIVE LOGITS
     ever
    1.12
     marginally
    0.99
     scratched
    0.85
     mildly
    0.83
     lasted
    0.80
     cared
    0.76
     vaguely
    0.75
     recently
    0.75
     spor
    0.74
     intermitt
    0.73
    Activations Density 0.304%

    No Known Activations