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

    Neuronpedia

    APICircuit 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. 11591
    Prev
    Next
    INDEX
    Explanations

    repetition or emphasis of the word "more."

    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
    501
    +0.12
    0.7%
    53
    +0.12
    0.7%
    338
    +0.12
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    501
    +0.12
    0.05
    239
    +0.12
    0.08
    284
    +0.12
    0.07
    Negative Logits
    ¼
    -3.70
    ´
    -3.26
    ¹
    -3.21
    ¾
    -3.16
    ĻĤ
    -3.16
    ·¸
    -3.15
    ½
    -3.04
    º
    -2.98
    ľĵ
    -2.97
    ·
    -2.92
    POSITIVE LOGITS
     than
    3.74
    Than
    3.26
    than
    2.95
     Than
    2.51
    judicial
    1.88
     manageable
    1.68
     forthcoming
    1.66
    likely
    1.59
     stringent
    1.58
     organised
    1.57
    Activations Density 0.199%

    No Known Activations