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. 12091
    Prev
    Next
    INDEX
    Explanations

    the name "Desire."

    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
    463
    +0.12
    0.7%
    46
    +0.12
    0.7%
    224
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    463
    +0.12
    0.02
    444
    +0.12
    0.01
    46
    +0.11
    0.01
    Negative Logits
    unnable
    -1.83
    ever
    -1.38
     proper
    -1.38
    %).
    -1.38
    ward
    -1.35
     behalf
    -1.35
    adem
    -1.34
    "){
    -1.33
    ifts
    -1.31
    aming
    -1.30
    POSITIVE LOGITS
    bourg
    1.80
     Rather
    1.75
    erialize
    1.72
    cribe
    1.69
    ²
    1.63
    irable
    1.62
    borough
    1.60
    criptions
    1.59
    perate
    1.58
    ilate
    1.57
    Activations Density 0.099%

    No Known Activations