© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    1. Home
    2. Under Peer Review · Sparse Autoencoders for Pythia-70M-Deduped
    3. Pythia-70M-Deduped
    4. Residuals
    5. 2-RES-SM
    6. 29863
    Prev
    Next
    INDEX
    Explanations

    phrases expressing an option or alternative

    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
    47
    +0.12
    0.7%
    428
    +0.10
    0.6%
    431
    +0.10
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    103
    +0.12
    0.02
    431
    +0.10
    0.02
    109
    +0.10
    0.02
    Negative Logits
    ľĵ
    -1.86
     importance
    -1.65
    oche
    -1.65
    []{
    -1.53
    '))
    -1.52
    atte
    -1.37
    ®
    -1.32
    ocks
    -1.31
    iste
    -1.31
    »¿
    -1.30
    POSITIVE LOGITS
    bard
    1.26
    pson
    1.25
     sworn
    1.23
     Category
    1.23
    holds
    1.23
    ktop
    1.22
     personally
    1.21
     done
    1.21
     directed
    1.20
    dorff
    1.20
    Activations Density 0.034%

    No Known Activations