© 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. 4-RES-JB
    11. 18530
    Prev
    Next
    INDEX
    Explanations

    contractions with negative connotations

    oai_token-act-pair · gpt-3.5-turbo

    instances of the phrase "can't"

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

    No Comments

    Negative Logits
     Appears
    -0.72
    amer
    -0.71
    imity
    -0.67
    interstitial
    -0.65
     TED
    -0.65
    our
    -0.62
    CNN
    -0.59
    æĥ
    -0.59
    estate
    -0.58
    DragonMagazine
    -0.58
    POSITIVE LOGITS
     afford
    1.00
     necessarily
    0.98
    icably
    0.87
    icable
    0.84
     imagine
    0.82
     bother
    0.80
     stomach
    0.80
     luck
    0.79
     discern
    0.78
    urtles
    0.77
    Activations Density 0.034%

    No Known Activations