© 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 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. 14252
    Prev
    Next
    INDEX
    Explanations

    contractions or possessive forms of the verb "to be."

    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
    deaux
    -0.14
    격
    -0.14
    ä¹ĭ
    -0.13
    éĺħ读次æķ°
    -0.13
    igor
    -0.13
     ÐĴеÑĢÑħов
    -0.13
    uya
    -0.13
    ocu
    -0.13
    иÑģÑĤ
    -0.12
    lessly
    -0.12
    POSITIVE LOGITS
     not
    0.26
     been
    0.25
     like
    0.24
     true
    0.21
     hard
    0.21
     amazing
    0.21
     worth
    0.19
    true
    0.19
    not
    0.19
     easier
    0.18
    Activations Density 0.157%

    No Known Activations