© 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 EvalsExportsinterp-engineNEWAPI 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. 8-RES-JB
    11. 1110
    Prev
    Next
    INDEX
    Explanations

    generic phrases indicating comparison or lack of significance

    oai_token-act-pair · gpt-3.5-turbo

    the phrase "not much," indicating a lack of substantial content or variety

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    8
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    yne
    -0.81
    izoph
    -0.77
    kus
    -0.74
    ilts
    -0.71
    ĪĴ
    -0.70
    YES
    -0.70
    otton
    -0.68
    initely
    -0.68
    han
    -0.67
    Ĥİ
    -0.66
    POSITIVE LOGITS
     anymore
    1.22
     whatsoever
    0.94
     consolation
    0.88
     else
    0.87
     nor
    0.84
     bothered
    0.84
     bother
    0.79
     noticeable
    0.78
     avail
    0.78
     surprises
    0.76
    Activations Density 0.089%

    No Known Activations