© 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. Julius Han · Gated SAE for Llama3-8B-Instruct
    5. Llama3-8B-IT
    7. Residual Stream
    9. 25-RES-JH
    11. 28752
    Prev
    Next
    INDEX
    Explanations

    frequent occurrences of the prepositions "in" and "from" in various contexts

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    Juliushanhanhan/llama-3-8b-it-res/blocks.25.hook_resid_post
    Features
    65,536
    Data Type
    float32
    Hook Name
    blocks.25.hook_resid_post
    Hook Layer
    25
    Architecture
    gated
    Context Size
    1,024
    Dataset
    Juliushanhanhan/openwebtext-1b-llama3-tokenized-cxt-1024
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    irst
    -0.18
    ello
    -0.17
    inkle
    -0.17
    ัวร
    -0.15
    alo
    -0.15
    ouses
    -0.15
    pering
    -0.15
    nk
    -0.14
    ãģĿãĤĮ
    -0.14
    agan
    -0.14
    POSITIVE LOGITS
     front
    0.30
    front
    0.21
     FRONT
    0.21
     plain
    0.21
    aus
    0.20
    -front
    0.19
     the
    0.17
     between
    0.17
    shall
    0.16
     frente
    0.16
    Activations Density 0.135%

    No Known Activations