© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Joseph Bloom · Initial Gemma-2B SAEs
    5. Gemma-2B
    7. Residual Stream
    9. 6-RES-JB
    11. 5218
    Prev
    Next
    INDEX
    Explanations

    negations and contrasts in text

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    jbloom/Gemma-2b-Residual-Stream-SAEs/gemma_2b_blocks.6.hook_resid_post_16384_anthropic_fast_lr
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    HuggingFaceFW/fineweb
    Features
    16,384
    Data Type
    torch.float32
    Hook Point
    blocks.6.hook_resid_post
    Architecture
    standard
    Context Size
    1,024
    Dataset
    HuggingFaceFW/fineweb
    Hook Point Layer
    6
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    50
    +0.21
    1.2%
    1262
    +0.15
    0.9%
    1974
    +0.12
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    1262
    +0.21
    0.11
    1974
    +0.15
    0.09
    950
    +0.12
    0.09
    Negative Logits
    <bos>
    -3.35
    ⓧ
    -0.99
    /***
    -0.75
     rejoin
    -0.71
    <?
    -0.69
    /*!
    -0.66
     disbur
    -0.65
     modernize
    -0.63
    
    -0.63
    //*/
    -0.62
    POSITIVE LOGITS
    Not
    0.94
     Not
    0.87
     véhic
    0.85
    yanto
    0.83
     NOT
    0.83
    postolic
    0.82
     déchir
    0.82
    gruntled
    0.82
    NOT
    0.82
    not
    0.81
    Activations Density 0.356%

    No Known Activations