© 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. Under Peer Review · Attention SAE Research Paper
    5. GPT2-Small
    7. Attention Out
    9. 10-ATT-KK
    11. 20164
    Prev
    Next
    INDEX
    Explanations

    phrases expressing surprise or disbelief

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 10-att-kk
    Configuration
    ckkissane/attn-saes-gpt2-small-all-layers/gpt2-small_L10_Hcat_z_lr1.20e-03_l11.30e+00_ds24576_bs4096_dc1.00e-05_rsanthropic_rie25000_nr4_v9.pt
    Prompts (Dashboard)
    36,864 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    24,576
    Data Type
    float32
    Hook Name
    blocks.10.attn.hook_z
    Hook Layer
    10
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Head Attr Weights
    0:0.05
    1:0.09
    2:0.03
    3:0.04
    4:0.27
    5:0.04
    6:0.05
    7:0.18
    8:0.02
    9:0.06
    10:0.04
    11:0.07
    Negative Logits
     Improve
    -2.60
    Requirements
    -2.38
     ►
    -2.36
    ¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯
    -2.32
    Neg
    -2.31
     imperfect
    -2.28
    uler
    -2.26
    bad
    -2.24
    Vert
    -2.24
    advant
    -2.22
    POSITIVE LOGITS
     surprise
    4.47
     shock
    4.44
     shocked
    4.24
     shocking
    4.08
     surprising
    4.03
     stunned
    4.02
     surprised
    3.82
     astonished
    3.81
     startling
    3.72
     bewild
    3.67
    Activations Density 0.075%

    No Known Activations