© 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. 3-ATT-KK
    11. 13615
    Prev
    Next
    INDEX
    Explanations

    justification or defense-related words or phrases

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot

    instances of justification or reasoning related to controversial actions and policies, particularly in a political or ethical context

    oai_token-act-pair · gpt-4o-miniTriggered 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
    Not in Any Lists

    No Comments

    Head Attr Weights
    0:0.02
    1:0.03
    2:0.08
    3:0.04
    4:0.02
    5:0.02
    6:0.05
    7:0.08
    8:0.22
    9:0.16
    10:0.06
    11:0.16
    Negative Logits
    washer
    -1.25
     brill
    -1.17
     Blues
    -1.14
     Emir
    -1.12
     Piet
    -1.10
     Lia
    -1.10
     prob
    -1.08
     fermented
    -1.08
     negotiator
    -1.07
     converter
    -1.06
    POSITIVE LOGITS
    reasonable
    1.46
    gging
    1.37
    usage
    1.34
    ooting
    1.33
     foregoing
    1.29
    dding
    1.29
    advertising
    1.28
    rosso
    1.27
    aggressive
    1.26
    SPONSORED
    1.25
    Activations Density 0.018%

    No Known Activations