© 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. 11-LLAMASCOPE-RES-32K
    11. 2305
    Prev
    Next
    INDEX
    Explanations

    phrases indicating significant material impact or damage

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With LLAMA3.1-8B @ 11-llamascope-res-32k
    Configuration
    fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L11R-8x
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    cerebras/SlimPajama-627B
    Features
    32,768
    Data Type
    bfloat16
    Hook Name
    blocks.11.hook_resid_post
    Hook Layer
    11
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    cerebras/SlimPajama-627B
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     _
    -0.15
    andr
    -0.15
     burial
    -0.15
     far
    -0.15
     geom
    -0.14
    oure
    -0.14
    izi
    -0.14
    far
    -0.13
     soon
    -0.13
    ancy
    -0.13
    POSITIVE LOGITS
    iola
    0.18
    ë§ī
    0.17
    svc
    0.16
    igua
    0.16
    ngine
    0.15
    827
    0.15
    enderit
    0.15
    ÙħÙĪÙĦ
    0.15
    acco
    0.14
    ulp
    0.14
    Activations Density 0.205%

    No Known Activations