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

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Under Peer Review · Sparse Autoencoders for Pythia-70M-Deduped
    5. Pythia-70M-Deduped
    7. Residuals
    9. 4-RES-SM
    11. 11707
    Prev
    Next
    INDEX
    Explanations

    numerical values or mathematical expressions related to calculations or sequences

    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/4-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.4.hook_resid_post
    Hook Layer
    4
    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
    21
    +0.20
    1.2%
    255
    +0.15
    0.9%
    413
    +0.11
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    255
    +0.20
    0.01
    84
    +0.15
    0.03
    448
    +0.11
    0.03
    Negative Logits
    »¿
    -2.38
    ĻĤ
    -2.33
    ĨĴ
    -2.10
    -1.85
    <|outofrange|>
    -1.85
    <|outofrange|>
    -1.85
    <|outofrange|>
    -1.85
    -1.85
    -1.85
    ↵↵
    -1.85
    POSITIVE LOGITS
    MHz
    1.68
    1451450014514500
    1.54
     reflections
    1.53
     MHz
    1.53
     times
    1.50
    00
    1.47
    ensity
    1.45
    irections
    1.44
    }_{-
    1.44
    YE
    1.40
    Activations Density 0.101%

    No Known Activations