© 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. 12055
    Prev
    Next
    INDEX
    Explanations

    phrases related to crime scenes and investigations

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

    references to specific locations or spatial contexts within narratives

    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
    2:0.07
    3:0.03
    4:0.03
    5:0.09
    6:0.07
    7:0.35
    8:0.05
    9:0.04
    10:0.10
    11:0.08
    Negative Logits
    ajor
    -1.43
    ahime
    -1.35
    ulative
    -1.25
    azeera
    -1.25
    jiang
    -1.25
     ecosystem
    -1.19
    etric
    -1.18
    ennial
    -1.17
    arget
    -1.13
    ocus
    -1.13
    POSITIVE LOGITS
    livious
    1.44
     unsuspecting
    1.33
    tyard
    1.33
    SPONSORED
    1.32
     squats
    1.26
     Defendant
    1.26
     srfAttach
    1.26
     desks
    1.24
     passers
    1.24
     makeshift
    1.21
    Activations Density 0.331%

    No Known Activations