© 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. Joseph Bloom · Initial Gemma-2B SAEs
    5. Gemma-2B
    7. Residual Stream
    9. 0-RES-JB
    11. 1805
    Prev
    Next
    INDEX
    Explanations

    recipe instructions and reviews related to cooking

    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.0.hook_resid_post_16384_anthropic
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    HuggingFaceFW/fineweb
    Features
    16,384
    Data Type
    torch.float32
    Hook Point
    blocks.0.hook_resid_post
    Architecture
    standard
    Context Size
    1,024
    Dataset
    HuggingFaceFW/fineweb
    Hook Point Layer
    0
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    1741
    +0.09
    0.3%
    752
    +0.07
    0.3%
    1120
    +0.07
    0.3%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    2044
    +0.09
    0.07
    1871
    +0.07
    0.03
    1080
    +0.07
    0.05
    Negative Logits
    ⓧ
    -0.83
    하세요
    -0.80
     sein
    -0.79
     continue
    -0.79
    public
    -0.78
     so
    -0.77
    |}
    -0.77
     do
    -0.76
     may
    -0.76
     also
    -0.76
    POSITIVE LOGITS
     affor
    2.29
     increa
    2.28
     reluct
    2.17
     stockholm
    2.17
     carrefour
    2.16
     disagre
    2.15
     impra
    2.15
     vété
    2.14
     shenan
    2.13
     quoique
    2.12
    Activations Density 0.375%

    No Known Activations