© 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. 1-RES-JB
    11. 10405
    Prev
    Next
    INDEX
    Explanations

    ancient or mystical looking symbols and names

    oai_token-act-pair · gpt-3.5-turbo

    multicharacter Unicode symbols and specific sequences impacted by encoding

    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
     dividing
    -0.63
    caps
    -0.63
     step
    -0.62
     wid
    -0.62
    eyes
    -0.60
     metast
    -0.59
    okin
    -0.58
    eller
    -0.57
     creep
    -0.57
     broker
    -0.57
    POSITIVE LOGITS
    Äģ
    1.33
    Ä«
    1.26
    Å«
    1.09
    Åį
    1.06
    á¹
    1.04
    á¸
    1.02
    Äĵ
    0.92
    ãĥ³ãĤ¸
    0.91
    ternity
    0.91
    ĥ
    0.88
    Activations Density 0.003%

    No Known Activations