© 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. Under Peer Review · Attention SAE Research Paper
    5. GPT2-Small
    7. Attention Out
    9. 5-ATT-KK
    11. 41971
    Prev
    Next
    INDEX
    Explanations

    occurrences of the phrase "come clean" and related expressions about honesty and transparency

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    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
    1:0.02
    2:0.08
    3:0.06
    4:0.11
    5:0.04
    6:0.03
    7:0.40
    8:0.03
    9:0.03
    10:0.07
    11:0.05
    Negative Logits
    lav
    -1.66
     Expansion
    -1.64
    pei
    -1.62
     Latvia
    -1.57
    growth
    -1.57
    isSpecialOrderable
    -1.54
    alt
    -1.49
     Liberties
    -1.49
    inav
    -1.48
    monary
    -1.48
    POSITIVE LOGITS
    DERR
    2.21
     disclosing
    1.99
    ELF
    1.97
    ��極
    1.88
    APD
    1.83
     wrongdoing
    1.74
    Reporting
    1.73
     factual
    1.72
     misled
    1.67
     culp
    1.67
    Activations Density 0.001%

    No Known Activations