© 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. 3-RES-JB
    11. 18146
    Prev
    Next
    INDEX
    Explanations

    specific nouns related to communication or correspondences such as "snippets", "messages", "tweets", "conversations", and "compliments"

    oai_token-act-pair · gpt-3.5-turbo

    references to written communication and related concepts

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 3-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    3
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    ded
    -0.65
    Args
    -0.63
    OV
    -0.63
    OUS
    -0.62
    shield
    -0.62
    plane
    -0.61
    lier
    -0.61
    ress
    -0.60
    Delivery
    -0.59
    fly
    -0.58
    POSITIVE LOGITS
    poons
    1.50
    mith
    1.44
    uggest
    1.38
    paces
    1.31
    pring
    1.28
    hops
    1.25
    hips
    1.24
    ettings
    1.23
    pace
    1.22
    ilver
    1.21
    Activations Density 0.150%

    No Known Activations