© 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
    3. Under Peer Review · Sparse Autoencoders for Pythia-70M-Deduped
    5. Pythia-70M-Deduped
    7. Residuals
    9. 2-RES-SM
    11. 11364
    Prev
    Next
    INDEX
    Explanations

    symbols or special characters in the document, particularly the character 'ł'

    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
    478
    +0.14
    0.8%
    110
    +0.13
    0.8%
    203
    +0.13
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    184
    +0.14
    0.01
    101
    +0.13
    0.01
    203
    +0.13
    0.01
    Negative Logits
    eless
    -1.78
    ensors
    -1.62
    ropolis
    -1.43
    Scan
    -1.39
     Forward
    -1.39
    ouch
    -1.39
    ibo
    -1.35
    odynamics
    -1.35
    ney
    -1.34
     Answer
    -1.34
    POSITIVE LOGITS
    ¼
    2.75
    Ĵ
    2.00
    ¸
    1.95
    ±
    1.86
    schild
    1.76
    «
    1.71
    į
    1.69
    IJ
    1.67
    ¾
    1.66
    ½
    1.64
    Activations Density 0.044%

    No Known Activations