© 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. 4-RES-JB
    11. 22091
    Prev
    Next
    INDEX
    Explanations

    references to the mountain named Mount Saint Helens, also known as Mount St. Helens

    oai_token-act-pair · gpt-3.5-turbo

    references to geographical locations and mountain names

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

    No Comments

    Negative Logits
     Hyundai
    -0.72
     MAC
    -0.66
     actionGroup
    -0.65
     reload
    -0.64
    abase
    -0.61
    redo
    -0.60
     subtract
    -0.60
    clave
    -0.60
     chamber
    -0.59
     attribution
    -0.59
    POSITIVE LOGITS
    burgh
    0.77
    ////////////////////////////////
    0.76
    ouf
    0.74
     Islands
    0.74
    eneg
    0.73
    kees
    0.69
    culosis
    0.68
    dden
    0.67
    LOD
    0.67
    çĶŁ
    0.66
    Activations Density 0.128%

    No Known Activations