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

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI 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. 27727
    Prev
    Next
    INDEX
    Explanations

    instances of user access permissions and related attributes within technical documents

    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
    263
    +0.21
    1.3%
    186
    +0.18
    1.1%
    23
    +0.17
    1.1%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    7
    +0.21
    0.02
    89
    +0.18
    0.08
    82
    +0.17
    0.07
    Negative Logits
    oire
    -1.57
    askell
    -1.43
    ader
    -1.40
    gradle
    -1.38
    sty
    -1.37
    aders
    -1.37
    OPLE
    -1.36
    ENE
    -1.36
    uler
    -1.36
    glass
    -1.35
    POSITIVE LOGITS
    >/
    1.97
    Īĺ
    1.54
    ?_
    1.54
    ¿½
    1.43
    »¿
    1.38
    rud
    1.33
     hone
    1.29
     excess
    1.29
     ranch
    1.28
    ĸ´
    1.28
    Activations Density 3.884%

    No Known Activations