© 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. 8336
    Prev
    Next
    INDEX
    Explanations

    mentions of specific geographic locations, particularly islands

    oai_token-act-pair · gpt-3.5-turbo

    references to specific islands and related geographical terms

    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
    nikov
    -0.98
    anova
    -0.83
    rss
    -0.79
    adal
    -0.77
     HF
    -0.75
     Maxim
    -0.73
    abal
    -0.73
     Rad
    -0.71
     Cav
    -0.70
    urgical
    -0.69
    POSITIVE LOGITS
     Isle
    2.72
     Poo
    1.82
    ËĪ
    1.59
     Cay
    1.50
     Donkey
    1.25
     enzyme
    1.18
     Kee
    1.16
     Isles
    1.15
     enzymes
    1.14
     Kau
    1.09
    Activations Density 0.072%

    No Known Activations