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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. OpenMOSS · Llama Scope: SAEs for Llama-3.1-8B
    3. Llama3.1-8B (Base)
    4. Residual Stream
    5. 23-LLAMASCOPE-RES-32K
    6. 15031
    Prev
    Next
    INDEX
    Explanations

    instances of the word "hang" and its variations, which implies a focus on social interactions or leisure activities

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With LLAMA3.1-8B @ 23-llamascope-res-32k
    Configuration
    fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L23R-8x
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    cerebras/SlimPajama-627B
    Features
    32,768
    Data Type
    bfloat16
    Hook Name
    blocks.23.hook_resid_post
    Hook Layer
    23
    Architecture
    jumprelu
    Context Size
    1,024
    Dataset
    cerebras/SlimPajama-627B
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    akedown
    -0.16
    fully
    -0.16
    /*/
    -0.16
    er
    -0.15
    kos
    -0.15
    pirit
    -0.14
    ño
    -0.14
     discern
    -0.14
     Ale
    -0.14
     Sylv
    -0.13
    POSITIVE LOGITS
     hang
    0.32
     Hang
    0.31
    Hang
    0.29
     Hanging
    0.29
     hangs
    0.28
     hanging
    0.28
    hang
    0.27
     hung
    0.25
    ry
    0.24
     Tough
    0.23
    Activations Density 0.015%

    No Known Activations