© 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. 3-RES-JB
    11. 23774
    Prev
    Next
    INDEX
    Explanations

    Hebrew characters and specific numeric sequences

    oai_token-act-pair · gpt-3.5-turbo

    Arabic or non-Latin characters and their patterns in the text

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 3-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    3
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    iko
    -0.83
    psey
    -0.82
    onge
    -0.81
    upiter
    -0.77
    oké
    -0.75
    ktop
    -0.74
    steen
    -0.74
    orget
    -0.72
    tten
    -0.72
    iphany
    -0.71
    POSITIVE LOGITS
    ÙĨ
    1.63
    Ùĩ
    1.62
    Ù
    1.61
    ا
    1.56
    د
    1.51
    اØ
    1.47
    Ø
    1.47
    Ùħ
    1.44
    ÙĬ
    1.43
    س
    1.43
    Activations Density 0.003%

    No Known Activations