© 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. OpenMOSS · Llama Scope: SAEs for Llama-3.1-8B
    5. Llama3.1-8B (Base)
    7. Residual Stream
    9. 18-LLAMASCOPE-RES-32K
    11. 26345
    Prev
    Next
    INDEX
    Explanations

    comas and various punctuation marks, indicating the neuron is focused on sentence structure and formatting in written text

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With LLAMA3.1-8B @ 18-llamascope-res-32k
    Configuration
    fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L18R-8x
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    cerebras/SlimPajama-627B
    Features
    32,768
    Data Type
    bfloat16
    Hook Name
    blocks.18.hook_resid_post
    Hook Layer
    18
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    cerebras/SlimPajama-627B
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     YYS
    -0.16
    pers
    -0.15
    respect
    -0.15
    stride
    -0.15
    burg
    -0.15
    len
    -0.15
    rather
    -0.15
    fang
    -0.14
    noho
    -0.14
    kol
    -0.14
    POSITIVE LOGITS
    ..
    0.17
    `
    0.15
    );↵
    0.14
    114
    0.14
    ab
    0.14
     Obr
    0.14
    |[
    0.13
    626
    0.13
    ì¹ľ
    0.13
    );
    0.13
    Activations Density 0.263%

    No Known Activations