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. 12-RES-JB
    6. 7766
    Prev
    Next
    INDEX
    Explanations

    punctuation and sentence endings

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    jbloom/Gemma-2b-Residual-Stream-SAEs/gemma_2b_blocks.12.hook_resid_post_16384
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    HuggingFaceFW/fineweb
    Features
    16,384
    Data Type
    torch.float32
    Hook Point
    blocks.12.hook_resid_post
    Architecture
    standard
    Context Size
    1,024
    Dataset
    HuggingFaceFW/fineweb
    Hook Point Layer
    12
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    50
    +0.40
    1.4%
    453
    +0.20
    0.7%
    2019
    +0.20
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    360
    +0.40
    0.05
    2019
    +0.20
    0.05
    543
    +0.20
    0.04
    Negative Logits
    <bos>
    -1.57
     Abbé
    -1.27
     Mlle
    -1.27
     tremb
    -1.22
     pamph
    -1.20
     suscep
    -1.20
     suspic
    -1.14
     fta
    -1.12
     hentai
    -1.10
     reluct
    -1.09
    POSITIVE LOGITS
    Discografia
    0.58
    MEMORANDUM
    0.56
    Tecnologia
    0.55
    CharField
    0.54
     وصلة
    0.53
    HKEY
    0.53
    存于互联网档案馆
    0.53
    Typeface
    0.52
    Historie
    0.52
    Timetable
    0.52
    Activations Density 0.207%

    No Known Activations