© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. OpenAI · Sparse Autoencoder for GPT2-Small - v5 2024
    5. GPT2-Small
    7. Residual Post - 32k
    9. 10-RES_POST_32K-OAI
    11. 28468
    Prev
    Next
    INDEX
    Explanations

    phrases emphasizing the importance of qualities or attributes, particularly using the article "a" followed by adjectives

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 10-res_post_32k-oai
    Configuration
    jbloom/GPT2-Small-OAI-v5-32k-resid-post-SAEs/v5_32k_layer_10.pt
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    32,768
    Data Type
    torch.float32
    Hook Name
    blocks.10.hook_resid_post
    Hook Layer
    10
    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
    425
    +0.15
    0.7%
    218
    +0.11
    0.5%
    564
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    36
    +0.15
    0.15
    250
    +0.11
    0.12
    159
    +0.10
    0.11
    Negative Logits
     Ruins
    -0.69
     Cous
    -0.68
    imov
    -0.63
    ãĤ´
    -0.63
     Ort
    -0.61
    ibrary
    -0.59
    ansk
    -0.58
     Lands
    -0.58
    ischer
    -0.58
    etz
    -0.58
    POSITIVE LOGITS
     truthful
    1.23
     cohesive
    1.20
     satisfactory
    1.19
     accurate
    1.18
     enjoyable
    1.18
     honest
    1.17
     acceptable
    1.16
     consistent
    1.15
     balanced
    1.15
     unbiased
    1.14
    Activations Density 0.686%

    No Known Activations