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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Under Peer Review · Sparse Autoencoders for Pythia-70M-Deduped
    3. Pythia-70M-Deduped
    4. Residuals
    5. 2-RES-SM
    6. 12074
    Prev
    Next
    INDEX
    Explanations

    the presence of various forms and mentions of the word "vibr."

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    ctigges/pythia-70m-deduped__res-sm_processed/2-res-sm
    Prompts (Dashboard)
    32,768 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    32,768
    Data Type
    torch.float32
    Hook Name
    blocks.2.hook_resid_post
    Hook Layer
    2
    Architecture
    standard
    Context Size
    128
    Dataset
    EleutherAI/the_pile_deduplicated
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    203
    +0.12
    0.7%
    74
    +0.10
    0.6%
    339
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    386
    +0.12
    0.01
    185
    +0.10
    0.01
    74
    +0.10
    0.02
    Negative Logits
    optera
    -1.95
     contrary
    -1.87
    GREEK
    -1.65
     intelligence
    -1.55
     laundering
    -1.52
    itoring
    -1.49
     whisper
    -1.49
     opposite
    -1.49
    $).
    -1.48
    onym
    -1.44
    POSITIVE LOGITS
    Ĵ
    2.87
    ĻĤ
    2.86
    ģ
    2.43
    ¦
    2.42
    ŀ
    2.40
    ij
    2.38
    ¨
    2.30
    ³
    2.24
    °
    2.23
    ¤
    2.15
    Activations Density 0.185%

    No Known Activations