© 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 EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    3. OpenMOSS · Llama Scope: SAEs for Llama-3.1-8B
    5. Llama3.1-8B (Base)
    7. Residual Stream
    9. 8-LLAMASCOPE-RES-32K
    11. 18541
    Prev
    Next
    INDEX
    Explanations

    phrases related to being trapped or stuck in various contexts

    oai_token-act-pair · gpt-4o-miniTriggered by @bot

    trapped

    np_max-act-logits · claude-4-5-sonnetTriggered by @sripadkarne
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With LLAMA3.1-8B @ 8-llamascope-res-32k
    Configuration
    fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L8R-8x
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    cerebras/SlimPajama-627B
    Features
    32,768
    Data Type
    bfloat16
    Hook Name
    blocks.8.hook_resid_post
    Hook Layer
    8
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    cerebras/SlimPajama-627B
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    ollapse
    -0.16
    HN
    -0.15
    imler
    -0.15
    stab
    -0.15
    epad
    -0.15
    .Formatting
    -0.14
    adan
    -0.14
    SSION
    -0.14
    351
    -0.13
    ollapsed
    -0.13
    POSITIVE LOGITS
    .ide
    0.17
    Ø©
    0.17
     Situation
    0.15
    aine
    0.15
    cheid
    0.15
    dealloc
    0.14
    arte
    0.13
    Ĥ¬
    0.13
    gaard
    0.13
    .squeeze
    0.13
    Activations Density 0.059%

    No Known Activations