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

    Neuronpedia

    APICircuit 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. 29564
    Prev
    Next
    INDEX
    Explanations

    elements and structures moving through physical or metaphorical constructions

    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
    156
    +0.18
    1.0%
    159
    +0.18
    1.0%
    34
    +0.13
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    159
    +0.18
    0.13
    73
    +0.18
    0.05
    277
    +0.13
    0.09
    Negative Logits
    olars
    -1.46
    quo
    -1.41
    urrent
    -1.38
     hostage
    -1.30
    alities
    -1.30
    onomy
    -1.29
    acted
    -1.28
    OSE
    -1.27
     transmitted
    -1.27
    .^\[[@
    -1.23
    POSITIVE LOGITS
     Studio
    1.56
     %{
    1.50
     fres
    1.44
     studio
    1.37
    ew
    1.36
     autom
    1.34
     closures
    1.33
     pipelines
    1.31
     Levi
    1.31
     engineers
    1.31
    Activations Density 5.182%

    No Known Activations