Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Under Peer Review · Attention SAE Research Paper
    3. GPT2-Small
    4. Attention Out
    5. 5-ATT-KK
    6. 19102
    Prev
    Next
    INDEX
    Explanations

    references to countries and their political interactions, particularly focusing on Russia, North Korea, Ukraine, and related geopolitical issues

    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
    0:0.06
    1:0.03
    2:0.03
    3:0.04
    4:0.03
    5:0.23
    6:0.01
    7:0.02
    8:0.04
    9:0.15
    10:0.16
    11:0.14
    Negative Logits
    sembly
    -1.49
    acebook
    -1.49
    anchester
    -1.46
    ournal
    -1.45
    idth
    -1.37
     Ambro
    -1.30
    types
    -1.29
    glomer
    -1.28
    gebra
    -1.26
    ngth
    -1.25
    POSITIVE LOGITS
     respons
    1.35
     refusal
    1.34
     willingness
    1.31
     insistence
    1.29
     retali
    1.26
     abst
    1.25
     acquies
    1.24
     goodwill
    1.19
     unwillingness
    1.18
     promptly
    1.15
    Activations Density 0.140%

    No Known Activations