© 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. 30793
    Prev
    Next
    INDEX
    Explanations

    phrases indicating future actions or obligations

    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
    581
    +0.11
    0.5%
    591
    +0.10
    0.4%
    156
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    551
    +0.11
    0.09
    49
    +0.10
    0.07
    139
    +0.10
    0.06
    Negative Logits
    otom
    -0.67
    oldemort
    -0.65
     aloud
    -0.64
     coerc
    -0.63
    ahead
    -0.62
    heimer
    -0.60
    Guard
    -0.60
    ividual
    -0.60
    aco
    -0.59
    arin
    -0.59
    POSITIVE LOGITS
     notice
    0.96
     want
    0.86
     LOVE
    0.86
     find
    0.84
     notices
    0.83
    noticed
    0.82
     liking
    0.79
     regret
    0.79
     noticed
    0.78
     Finding
    0.77
    Activations Density 0.363%

    No Known Activations