© 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. 8103
    Prev
    Next
    INDEX
    Explanations

    conjunctions and contextually significant phrases related to balance or duality

    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
    314
    +0.12
    0.5%
    374
    +0.11
    0.5%
    533
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    393
    +0.12
    0.08
    678
    +0.11
    0.06
    746
    +0.11
    0.03
    Negative Logits
     Flavoring
    -0.75
    oulos
    -0.71
    manship
    -0.70
     Cosponsors
    -0.70
     Scient
    -0.69
     Scientology
    -0.69
     Volunte
    -0.69
     Lutheran
    -0.67
     Proposition
    -0.66
     ..............
    -0.66
    POSITIVE LOGITS
     bottom
    1.16
     periphery
    1.10
     upper
    1.05
     underside
    1.04
     aft
    0.98
     attic
    0.98
     sides
    0.97
     rear
    0.96
    middle
    0.96
     roof
    0.95
    Activations Density 1.018%

    No Known Activations