© 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. 19145
    Prev
    Next
    INDEX
    Explanations

    mentions of the word "up"

    oai_token-act-pair · gpt-3.5-turbo

    repeated mentions of the word "puppet."

    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
     recorded
    -0.66
     Riot
    -0.65
     Elemental
    -0.65
     Catalyst
    -0.63
     drums
    -0.63
     Realm
    -0.61
     Tracks
    -0.61
     badges
    -0.61
     Curse
    -0.60
     Crusader
    -0.60
    POSITIVE LOGITS
    upp
    4.67
    uppet
    1.88
    upper
    1.49
    ups
    1.20
    utt
    1.15
    up
    1.15
    supp
    1.13
     Pupp
    1.11
    anty
    0.99
    opp
    0.99
    Activations Density 0.008%

    No Known Activations