© 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. Llama3.3-70B-IT
    5. 50-RESID-POST-GF
    7. 40300
    Prev
    Next
    INDEX
    Explanations

    numbers and codes

    np_acts-logits-general · gemini-2.5-flash-lite

    structural and metadata delimiters in text—role headers, list numbering, and separator punctuation like colons/periods/underscores that mark boundaries.

    oai_token-act-pair · gpt-5Triggered by @vetterc0
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With LLAMA3.3-70B-IT @ 50-resid-post-gf
    Configuration
    Goodfire/Llama-3.3-70B-Instruct-SAE-l50/Llama-3.3-70B-Instruct-SAE-l50.pt
    Prompts (Dashboard)
    10,000 prompts, 128 tokens each
    Dataset (Dashboard)
    lmsys/lmsys-chat-1m
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    ******č\n
    -0.11
    .Formatter
    -0.10
    okedex
    -0.10
    EMPLARY
    -0.10
    ÂĢÂĢ
    -0.10
    ¶Į
    -0.10
    įng
    -0.09
    yonel
    -0.09
    hiba
    -0.09
    ıa
    -0.09
    POSITIVE LOGITS
    彡
    0.09
     Bro
    0.08
     Organ
    0.08
    cl
    0.08
     Haus
    0.08
     Dah
    0.08
    bal
    0.08
     lim
    0.08
    erialized
    0.08
    af
    0.07
    Activations Density 0.293%

    No Known Activations