© 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. OpenMOSS · Llama Scope: SAEs for Llama-3.1-8B
    5. Llama3.1-8B (Base)
    7. Residual Stream
    9. 2-LLAMASCOPE-RES-32K
    11. 6973
    Prev
    Next
    INDEX
    Explanations

    mentions of geographical locations, particularly in South America

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With LLAMA3.1-8B @ 2-llamascope-res-32k
    Configuration
    fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L2R-8x
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    cerebras/SlimPajama-627B
    Features
    32,768
    Data Type
    bfloat16
    Hook Name
    blocks.2.hook_resid_post
    Hook Layer
    2
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    cerebras/SlimPajama-627B
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    207
    -0.14
    ua
    -0.14
    bia
    -0.14
    udo
    -0.14
    208
    -0.14
    Æ¡
    -0.13
     Huff
    -0.13
    agas
    -0.13
     pub
    -0.13
    rane
    -0.13
    POSITIVE LOGITS
    anto
    0.17
    omy
    0.17
    ignum
    0.14
    warts
    0.14
     Wil
    0.14
     localVar
    0.14
    oppel
    0.14
    oose
    0.14
    htmlspecialchars
    0.13
    nem
    0.13
    Activations Density 0.052%

    No Known Activations