© 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. 1872
    Prev
    Next
    INDEX
    Explanations

    references to weapons of mass destruction and instances of extensive damage or devastation

    oai_token-act-pair · gpt-3.5-turbo

    phrases related to destruction

    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
    edin
    -0.89
    yi
    -0.82
     Asked
    -0.77
    igh
    -0.76
    liam
    -0.72
     fielded
    -0.67
    ®
    -0.67
     BMI
    -0.66
    IRC
    -0.65
     ®
    -0.65
    POSITIVE LOGITS
     destruction
    3.62
     Destruction
    2.38
     devastation
    2.20
     annihilation
    2.06
     destroying
    1.87
     demolition
    1.85
     destroy
    1.70
     carnage
    1.69
     destroyed
    1.68
     destruct
    1.61
    Activations Density 0.020%

    No Known Activations