© 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. 3-RES-JB
    11. 17650
    Prev
    Next
    INDEX
    Explanations

    The neuron is primarily looking for sentences containing the word "cannot" followed by a verb

    oai_token-act-pair · gpt-3.5-turbo

    phrases indicating inability or impossibility

    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
    bats
    -0.67
     Generation
    -0.65
    flame
    -0.65
    grounds
    -0.64
    ItemThumbnailImage
    -0.62
    ItemImage
    -0.62
     classy
    -0.61
    pocket
    -0.61
    scene
    -0.61
    soDeliveryDate
    -0.60
    POSITIVE LOGITS
     necessarily
    1.03
     afford
    0.99
    adian
    0.93
    't
    0.93
    berra
    0.89
     reproduce
    0.88
    icably
    0.83
    urtles
    0.82
     exceed
    0.82
     easily
    0.81
    Activations Density 0.043%

    No Known Activations