© 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. 0-RES-JB
    11. 19872
    Prev
    Next
    INDEX
    Explanations

    references to titles or characters related to authority or power, such as titles like "Overlord" or "Drug Lord"

    oai_token-act-pair · gpt-3.5-turbo

    repeated mentions of the term "lord" in various contexts, particularly in relation to power and authority

    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
     sequ
    -0.73
     canv
    -0.69
     similarly
    -0.66
     declass
    -0.65
     digitally
    -0.64
     sharp
    -0.61
     Fresh
    -0.60
     SAF
    -0.60
     basis
    -0.58
     Disapp
    -0.58
    POSITIVE LOGITS
    lord
    4.56
    lords
    3.33
     lord
    1.67
    Lord
    1.43
     lords
    1.32
    lore
    1.27
    boss
    1.19
     Lord
    1.08
    hawk
    1.04
    mare
    1.02
    Activations Density 0.015%

    No Known Activations