© 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. 3-ATT-KK
    11. 12520
    Prev
    Next
    INDEX
    Explanations

    mentions of numbers and statistics in a factual context

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

    punctuation marks and their associated contexts in text

    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
    1:0.07
    2:0.08
    3:0.03
    4:0.05
    5:0.11
    6:0.10
    7:0.07
    8:0.08
    9:0.16
    10:0.11
    11:0.07
    Negative Logits
    together
    -1.35
    kered
    -1.20
    ependent
    -1.16
     unite
    -1.11
     ourselves
    -1.11
     realize
    -1.10
    oder
    -1.07
     parted
    -1.07
     rejoice
    -1.07
    tern
    -1.06
    POSITIVE LOGITS
     ingred
    1.31
     cannabin
    1.27
     (£
    1.22
    opsis
    1.22
    ��
    1.18
     Compass
    1.16
     "…
    1.12
    OTOS
    1.10
     Trog
    1.07
     Rept
    1.06
    Activations Density 0.050%

    No Known Activations