© 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. 2430
    Prev
    Next
    INDEX
    Explanations

    the word "rah" with a high activation value

    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
    1664
    +0.07
    0.2%
    233
    +0.06
    0.2%
    1003
    +0.06
    0.2%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    283
    +0.07
    0.02
    1343
    +0.06
    0.05
    1927
    +0.06
    0.04
    Negative Logits
     knocked
    -0.64
     hard
    -0.60
     job
    -0.60
     turned
    -0.58
     forced
    -0.58
     missed
    -0.58
     potrzebu
    -0.57
     main
    -0.57
     advanced
    -0.57
     repeated
    -0.57
    POSITIVE LOGITS
    rah
    2.25
    RAH
    1.53
     grati
    1.49
     socie
    1.41
     dispen
    1.40
     encre
    1.35
     oner
    1.34
     bourgeo
    1.34
     pecuni
    1.34
     frau
    1.34
    Activations Density 0.358%

    No Known Activations