Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    © Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Julius Han · Gated SAE for Llama3-8B-Instruct
    3. Llama3-8B-IT
    4. Residual Stream
    5. 25-RES-JH
    6. 9253
    Prev
    Next
    INDEX
    Explanations

    features related to products and their specifications

    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
    ayer
    -0.16
    lean
    -0.15
    uje
    -0.14
     Ñĥм
    -0.14
     tiers
    -0.14
    enga
    -0.14
    deck
    -0.14
    ep
    -0.13
     laps
    -0.13
    anh
    -0.13
    POSITIVE LOGITS
     carrying
    0.26
     instruction
    0.24
     Instruction
    0.22
     instructional
    0.21
     carry
    0.20
     instructions
    0.20
     included
    0.19
     bonus
    0.19
     Carry
    0.19
    Instruction
    0.19
    Activations Density 0.068%

    No Known Activations