© 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. 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. 13550
    Prev
    Next
    INDEX
    Explanations

    words related to causing harm or damage

    oai_token-act-pair · gpt-3.5-turbo

    phrases related to causing harm or damage

    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
    abal
    -0.88
    uates
    -0.82
    iamond
    -0.74
    anan
    -0.74
    olf
    -0.72
    uated
    -0.71
    liga
    -0.71
    imating
    -0.70
    imentary
    -0.69
    assador
    -0.68
    POSITIVE LOGITS
    GBT
    0.70
     Mankind
    0.67
    Instead
    0.63
    Redditor
    0.63
    ãĥ³ãĤ¸
    0.62
    Ult
    0.62
     incompet
    0.62
    ppe
    0.61
    Ger
    0.61
     sanity
    0.61
    Activations Density 0.000%

    No Known Activations