© 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. 11250
    Prev
    Next
    INDEX
    Explanations

    key features and descriptors related to design, particularly in products and materials

    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
    326
    +0.20
    1.1%
    153
    +0.15
    0.9%
    198
    +0.15
    0.9%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    140
    +0.20
    0.10
    326
    +0.15
    0.07
    68
    +0.15
    0.11
    Negative Logits
     referendum
    -1.54
     Emirates
    -1.52
     Ministry
    -1.51
     Brussels
    -1.50
     team
    -1.50
     voting
    -1.45
     notation
    -1.43
     refe
    -1.40
     memorandum
    -1.39
     decision
    -1.38
    POSITIVE LOGITS
    ingly
    1.70
     lives
    1.55
    adelphia
    1.54
    itness
    1.48
    thening
    1.46
    Ĺ
    1.43
    ~),
    1.41
     pores
    1.41
    sky
    1.40
    uh
    1.39
    Activations Density 4.623%

    No Known Activations