Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Joseph Bloom · Initial Gemma-2B SAEs
    3. Gemma-2B
    4. Residual Stream
    5. 6-RES-JB
    6. 5573
    Prev
    Next
    INDEX
    Explanations

    words and phrases related to technical processes and methodologies, such as "Lean Thinking," "Agile," "Lean Startup," "Six Sigma," "ERP," and "portfolio management."

    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
    +0.27
    1.1%
    1870
    +0.14
    0.5%
    1403
    +0.11
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    1870
    +0.27
    0.05
    1403
    +0.14
    0.04
    1385
    +0.11
    0.10
    Negative Logits
    <bos>
    -1.91
     intersper
    -1.61
     guarante
    -1.48
     encomp
    -1.45
     poff
    -1.29
     fta
    -1.29
     gild
    -1.28
     squa
    -1.28
     volunte
    -1.26
     wherea
    -1.26
    POSITIVE LOGITS
     Management
    0.75
     management
    0.75
     sarili
    0.73
    stdc
    0.69
     Bekasi
    0.67
     bagay
    0.67
     galeria
    0.67
     psicologia
    0.65
     hotelu
    0.65
     managers
    0.65
    Activations Density 1.831%

    No Known Activations