© 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. 7-RES-JB
    11. 12572
    Prev
    Next
    INDEX
    Explanations

    mentions of historical events or changes over time

    oai_token-act-pair · gpt-3.5-turbo

    references to historical context or past events

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 7-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    7
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    wake
    -0.74
    Redditor
    -0.71
    Dialogue
    -0.70
    =>
    -0.69
    Notes
    -0.68
    apply
    -0.67
    akens
    -0.67
    alias
    -0.65
     interven
    -0.64
    ibliography
    -0.64
    POSITIVE LOGITS
     existed
    0.73
     unthinkable
    0.70
     ruled
    0.70
     dinosaurs
    0.66
     dated
    0.66
     feared
    0.66
     linear
    0.63
     Linear
    0.63
     envisioned
    0.62
     innocence
    0.61
    Activations Density 1.273%

    No Known Activations