© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    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. 8-RES_POST_32K-OAI
    11. 5924
    Prev
    Next
    INDEX
    Explanations

    references to physical interaction or connection with objects and events

    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
    260
    +0.09
    0.4%
    46
    +0.09
    0.4%
    96
    +0.08
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    587
    +0.09
    0.06
    569
    +0.09
    0.06
    323
    +0.08
    0.06
    Negative Logits
    MQ
    -0.80
    Independent
    -0.75
    IRC
    -0.72
    ORTS
    -0.72
    Recomm
    -0.72
    VB
    -0.68
    HR
    -0.67
    HM
    -0.66
    taboola
    -0.64
    SHARE
    -0.63
    POSITIVE LOGITS
     bushes
    1.03
     flesh
    0.97
     cloth
    0.91
     walls
    0.91
     surfaces
    0.89
     rocks
    0.89
     skin
    0.87
     vegetation
    0.85
     strawberries
    0.81
     asphalt
    0.80
    Activations Density 0.532%

    No Known Activations