© 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. 15801
    Prev
    Next
    INDEX
    Explanations

    specific maneuvers or actions

    oai_token-act-pair · gpt-3.5-turbo

    variations of the word "refer" and its associated forms

    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
    alien
    -0.84
    olen
    -0.80
     neut
    -0.74
    umenthal
    -0.72
    unte
    -0.70
    oln
    -0.69
    enty
    -0.67
     unden
    -0.67
    reat
    -0.65
    icter
    -0.65
    POSITIVE LOGITS
    train
    0.78
    puff
    0.78
    borough
    0.77
    pole
    0.77
    ment
    0.75
     Barker
    0.73
    VILLE
    0.71
    ball
    0.70
    heads
    0.69
    ttes
    0.69
    Activations Density 0.215%

    No Known Activations