© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    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. 7733
    Prev
    Next
    INDEX
    Explanations

    words related to observation or surveillance

    oai_token-act-pair · gpt-3.5-turbo

    instances of the word "watching."

    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
    DN
    -0.68
    Orig
    -0.65
    erto
    -0.64
     CV
    -0.62
    Address
    -0.62
    ITION
    -0.62
    igen
    -0.61
    ilde
    -0.61
    ipient
    -0.60
    Internal
    -0.59
    POSITIVE LOGITS
     watching
    3.44
    watching
    2.50
     Watching
    2.49
     listening
    2.05
     watched
    1.99
     viewing
    1.95
     watch
    1.92
     witnessing
    1.84
     observing
    1.80
     watches
    1.80
    Activations Density 0.017%

    No Known Activations