© 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. 5845
    Prev
    Next
    INDEX
    Explanations

    variations of the verb "to be" in various conjugations

    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
    212
    +0.14
    0.6%
    566
    +0.12
    0.6%
    36
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    226
    +0.14
    0.13
    293
    +0.12
    0.12
    697
    +0.11
    0.10
    Negative Logits
    odor
    -0.72
    osion
    -0.71
     Levant
    -0.66
    WAR
    -0.64
    inav
    -0.64
    cius
    -0.63
    ITION
    -0.61
    isc
    -0.61
     Orchestra
    -0.60
     Presence
    -0.59
    POSITIVE LOGITS
     gonna
    1.24
     gotten
    1.10
     gotta
    1.10
    been
    1.05
     been
    1.00
     gone
    0.99
     done
    0.93
     got
    0.91
     forgot
    0.89
     messed
    0.88
    Activations Density 1.184%

    No Known Activations