© 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
    2. OpenAI · Sparse Autoencoder for GPT2-Small - v5 2024
    3. GPT2-Small
    4. Residual Post - 32k
    5. 8-RES_POST_32K-OAI
    6. 6443
    Prev
    Next
    INDEX
    Explanations

    references to The New York Times and its newsletters

    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
    373
    +0.11
    0.5%
    727
    +0.11
    0.5%
    246
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    373
    +0.11
    -0.00
    257
    +0.11
    0.02
    211
    +0.10
    0.01
    Negative Logits
    abase
    -0.69
     corps
    -0.69
     tatt
    -0.69
    inker
    -0.64
     neighb
    -0.62
     bun
    -0.62
     improvis
    -0.61
     scales
    -0.60
     fingert
    -0.60
     hobby
    -0.58
    POSITIVE LOGITS
    govtrack
    0.86
    MSN
    0.75
     Terms
    0.71
    ahoo
    0.71
    Privacy
    0.71
    Quotes
    0.68
     newsletters
    0.67
    ussions
    0.66
    cdn
    0.65
    isodes
    0.65
    Activations Density 0.055%

    No Known Activations