© 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. 7390
    Prev
    Next
    INDEX
    Explanations

    words that indicate perception or speculation

    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
    +0.14
    0.6%
    525
    +0.12
    0.5%
    118
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    195
    +0.14
    0.05
    8
    +0.12
    0.05
    561
    +0.11
    0.05
    Negative Logits
    owan
    -0.71
    cial
    -0.68
     heights
    -0.62
    ipop
    -0.59
    ocaust
    -0.58
    lift
    -0.58
    "},{"
    -0.58
    division
    -0.57
    eph
    -0.56
    raphics
    -0.56
    POSITIVE LOGITS
     likely
    1.01
     unlikely
    0.92
     destined
    0.90
     poised
    0.88
    é¾į
    0.88
     intent
    0.84
    likely
    0.84
     primed
    0.72
     willing
    0.72
     ideally
    0.71
    Activations Density 0.265%

    No Known Activations