© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI 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. 8-RES-JB
    11. 1159
    Prev
    Next
    INDEX
    Explanations

    words related to various items or concepts mentioned in a document, such as Medicare, Medicaid, Social Security, handguns, rifles, shotguns, gasoline, distillates, crude oil, rice, wheat, corn, salt, wool, etc

    oai_token-act-pair · gpt-3.5-turbo

    instances of conjunctions, specifically the word "and"

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    8
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    Redditor
    -0.74
    RM
    -0.71
    rb
    -0.70
    itor
    -0.69
    noon
    -0.67
    235
    -0.64
    :(
    -0.64
    :[
    -0.63
    :]
    -0.61
    tera
    -0.61
    POSITIVE LOGITS
     assorted
    1.13
    etc
    1.08
     etc
    0.89
     other
    0.84
     others
    0.83
    whatever
    0.82
    ĪĴ
    0.79
     downright
    0.76
     even
    0.75
     possibly
    0.75
    Activations Density 0.195%

    No Known Activations