© 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. 5-ATT-KK
    11. 424
    Prev
    Next
    INDEX
    Explanations

    phrases that suggest competition or head-to-head matchups

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 5-att-kk
    Configuration
    ckkissane/attn-saes-gpt2-small-all-layers/gpt2-small_L5_Hcat_z_lr1.20e-03_l11.00e+00_ds49152_bs4096_dc1.00e-06_rsanthropic_rie25000_nr4_v9.pt
    Prompts (Dashboard)
    36,864 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    49,152
    Data Type
    float32
    Hook Name
    blocks.5.attn.hook_z
    Hook Layer
    5
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Head Attr Weights
    0:0.02
    1:0.01
    2:0.11
    3:0.05
    4:0.17
    5:0.06
    6:0.03
    7:0.31
    8:0.04
    9:0.04
    10:0.07
    11:0.05
    Negative Logits
    ulic
    -1.67
    roud
    -1.58
    rh
    -1.53
    hered
    -1.50
    reon
    -1.46
    )</
    -1.45
    RIP
    -1.44
    iour
    -1.44
    potion
    -1.41
     Archives
    -1.37
    POSITIVE LOGITS
    ascus
    1.65
     competing
    1.59
     challengers
    1.59
     competitors
    1.53
     hitters
    1.51
    Winner
    1.51
     competition
    1.50
     counterparts
    1.49
    UGC
    1.48
     vying
    1.48
    Activations Density 0.003%

    No Known Activations