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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports 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. 1-RES-JB
    11. 14064
    Prev
    Next
    INDEX
    Explanations

    numeric values related to power consumption in watts

    oai_token-act-pair · gpt-3.5-turbo

    references to "Watt" and variations of "watt" in the context of power consumption or generation

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 1-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    1
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     parole
    -0.72
     Phantom
    -0.70
     Camb
    -0.64
    PsyNetMessage
    -0.63
     probation
    -0.62
     subsistence
    -0.62
     Cambodia
    -0.61
     Cambod
    -0.61
     Democracy
    -0.61
    ournal
    -0.60
    POSITIVE LOGITS
     Watt
    1.27
    atts
    1.04
    aday
    0.96
     watt
    0.94
    ynes
    0.85
    agers
    0.85
    borgh
    0.84
    acre
    0.84
    ucci
    0.84
    enberg
    0.84
    Activations Density 0.004%

    No Known Activations