© 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
    2. Qwen3-4B
    3. 23-TRANSCODER-HP
    4. 24510
    Prev
    Next
    INDEX
    Explanations

    Grammatical context

    np_max-act · gemini-2.0-flash

    strong contrasts (either through conjunctions like "but" or through conceptual/medical opposites) that indicate a divergence from expectation or norm.

    oai_token-act-pair · deepseek-v3Triggered by @finnwit

    Common connecting words that precede contrasting information.

    eleuther_acts_top20 · deepseek-v3Triggered by @finnwit
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    mwhanna/qwen3-4b-transcoders/layer_23.safetensors
    Prompts (Dashboard)
    16,384 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    163,840
    Data Type
    float32
    Hook Name
    blocks.23.mlp.hook_in
    Architecture
    transcoder
    Context Size
    8,192
    Dataset
    monology/pile-uncopyrighted
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    è¾ĥå¤ļ
    -0.26
    宣
    -0.26
    åłī
    -0.26
    æ£ī
    -0.25
    èĥ½åģļåΰ
    -0.24
    Playable
    -0.24
    èĥ½è¾¾åΰ
    -0.24
    Textarea
    -0.23
     ban
    -0.23
    è¿ĩå¤ļ
    -0.23
    POSITIVE LOGITS
    å¹¶ä¸įä¼ļ
    0.33
    并没æľī
    0.32
     nothing
    0.31
    æł¹æľ¬æ²¡æľī
    0.29
    ä¸įä¼ļæľī
    0.29
    å¹¶ä¸į
    0.29
    Rew
    0.28
    å¹¶æĹł
    0.28
    ä»İæĿ¥ä¸į
    0.28
    ])[
    0.28
    Activations Density 0.001%

    No Known Activations