© 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. 1570
    Prev
    Next
    INDEX
    Explanations

    proper names or terms related to the city of Edinburgh

    oai_token-act-pair · gpt-3.5-turbo

    the name "Ed" and its variations in different contexts

    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
     range
    -0.72
     Mafia
    -0.69
    azer
    -0.68
     squash
    -0.67
     tha
    -0.63
     distance
    -0.63
     LOT
    -0.62
     carrier
    -0.62
     proto
    -0.62
     ceasefire
    -0.62
    POSITIVE LOGITS
    Ed
    4.04
     Ed
    1.96
     ed
    1.43
     Edition
    1.34
    Edward
    1.28
    eds
    1.24
     Edison
    1.23
    Ell
    1.15
     Edwards
    1.15
    Roy
    1.12
    Activations Density 0.013%

    No Known Activations