© 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. 9-RES-JB
    11. 11218
    Prev
    Next
    INDEX
    Explanations

    contractions where the word 'not' is followed by an action verb

    oai_token-act-pair · gpt-3.5-turbo

    the usage of the conditional negative form "wouldn't"

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 9-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    9
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     mutually
    -0.63
     Gutenberg
    -0.60
     populated
    -0.60
     medi
    -0.60
     voic
    -0.59
     disabled
    -0.58
     nearest
    -0.58
     indexed
    -0.58
     misdem
    -0.56
     transmitted
    -0.56
    POSITIVE LOGITS
    't
    1.50
    n
    0.98
    itely
    0.95
    uts
    0.94
    ÃŃ
    0.93
    ´
    0.91
    gered
    0.89
    emonium
    0.88
    ÃĥÃĤ
    0.86
    terness
    0.85
    Activations Density 0.022%

    No Known Activations