© 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. 16698
    Prev
    Next
    INDEX
    Explanations

    mentions of different sections or categories within a text or document

    oai_token-act-pair · gpt-3.5-turbo

    references to sections within a text or document

    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
    urrencies
    -0.71
    天
    -0.66
    iour
    -0.65
    ÙĨ
    -0.61
    ilon
    -0.60
    KO
    -0.59
    ãĥķãĤ©
    -0.59
    IJ
    -0.59
    Ïī
    -0.59
    duino
    -0.58
    POSITIVE LOGITS
     section
    3.60
     sections
    2.81
    section
    2.37
     Section
    2.31
     subsection
    2.25
    Section
    2.15
     Sections
    2.04
    sections
    1.86
     subsections
    1.83
     SECTION
    1.74
    Activations Density 0.012%

    No Known Activations