© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW Community BlogPrivacy & TermsContact
    1. Home
    3. Under Peer Review · Attention SAE Research Paper
    5. GPT2-Small
    7. Attention Out
    9. 3-ATT-KK
    11. 14477
    Prev
    Next
    INDEX
    Explanations

    phrases related to crime scenes and legal/police investigative terms

    oai_token-act-pair · gpt-3.5-turbo
    Triggered by @bot

    incidents involving firearms and injuries or fatalities

    oai_token-act-pair · gpt-4o-mini
    Triggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 3-att-kk
    Configuration
    ckkissane/attn-saes-gpt2-small-all-layers/gpt2-small_L3_Hcat_z_lr1.20e-03_l19.00e-01_ds24576_bs4096_dc1.00e-06_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.3.attn.hook_z
    Hook Layer
    3
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Head Attr Weights
    0:0.03
    1:0.06
    2:0.07
    3:0.05
    4:0.03
    5:0.07
    6:0.05
    7:0.09
    8:0.18
    9:0.15
    10:0.07
    11:0.10
    Negative Logits
     Inher
    -1.53
     Critics
    -1.51
     introduce
    -1.38
     Invention
    -1.29
    global
    -1.27
    ゴン
    -1.24
    AFTA
    -1.24
     globalization
    -1.24
     Generation
    -1.24
     coined
    -1.23
    POSITIVE LOGITS
    anchez
    1.49
     incrim
    1.49
     suspicious
    1.41
     deserted
    1.39
     wreckage
    1.33
     unlawfully
    1.33
     zip
    1.31
     unfit
    1.31
    susp
    1.28
     charred
    1.26
    Activations Density 0.049%

    No Known Activations