© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Joseph Bloom · Open Source Sparse Autoencoders for all Residual Stream Layers of GPT2-Small
    5. GPT2-Small
    7. Residual Stream
    9. 0-RES-JB
    11. 12223
    Prev
    Next
    INDEX
    Explanations

    names related to different ethnic origins

    oai_token-act-pair · gpt-3.5-turbo

    references to specific nationalities or ethnic groups, particularly Swedish and Kiwi

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 0-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.hook_resid_pre
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    24,576
    Data Type
    torch.float32
    Hook Point
    blocks.0.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    0
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     Thur
    -0.86
    BAT
    -0.79
    MAL
    -0.78
    ï
    -0.75
     Disc
    -0.74
    Golden
    -0.72
    ea
    -0.72
     Var
    -0.72
    Ïģ
    -0.71
     AH
    -0.70
    POSITIVE LOGITS
     Swed
    2.58
     Kiw
    1.43
     Gomez
    1.37
    park
    1.19
     pes
    1.17
    sie
    1.15
     annexation
    1.15
     Sloven
    1.13
     Pes
    1.13
     Schne
    1.02
    Activations Density 0.080%

    No Known Activations