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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & 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. 12185
    Prev
    Next
    INDEX
    Explanations

    elements related to programming function definitions

    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
    16
    +0.17
    0.9%
    249
    +0.14
    0.8%
    115
    +0.13
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    249
    +0.17
    0.03
    112
    +0.14
    0.03
    16
    +0.13
    0.03
    Negative Logits
    imore
    -1.85
    illary
    -1.67
    ags
    -1.61
    oney
    -1.57
    oun
    -1.50
    aya
    -1.50
    ault
    -1.49
    ieri
    -1.45
    hang
    -1.45
    ht
    -1.44
    POSITIVE LOGITS
    Ĥ
    3.72
    ¬
    3.34
    ĥ½
    3.23
    ¶
    3.23
    ¦
    3.15
    ¸
    3.14
    İ
    3.13
    ¯
    3.12
    ĥ
    3.11
    ¤
    3.10
    Activations Density 0.143%

    No Known Activations