© 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. 4483
    Prev
    Next
    INDEX
    Explanations

    references to the word "Cer" and its variations, likely indicating it focuses on names or terms associated with the specific location or topic related to "Cer"

    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
    23
    +0.11
    0.5%
    92
    +0.10
    0.4%
    412
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    232
    +0.11
    0.04
    284
    +0.10
    0.03
    503
    +0.09
    0.02
    Negative Logits
    xual
    -0.65
    ership
    -0.65
    urst
    -0.64
     Cascade
    -0.62
     Lovecraft
    -0.61
    ¡
    -0.60
     Pres
    -0.57
    è£ħ
    -0.57
    ħĭ
    -0.57
    IGHTS
    -0.56
    POSITIVE LOGITS
    berus
    1.16
    rier
    1.06
    mination
    1.06
    amic
    1.04
    rera
    1.01
    rible
    1.01
    rigan
    0.99
    itage
    0.98
    sei
    0.97
    bal
    0.96
    Activations Density 0.098%

    No Known Activations