© 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. 13960
    Prev
    Next
    INDEX
    Explanations

    negative evaluations or descriptors of films, situations, or experiences

    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
    88
    +0.13
    0.7%
    360
    +0.12
    0.7%
    188
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    303
    +0.13
    0.03
    99
    +0.12
    0.03
    360
    +0.11
    0.02
    Negative Logits
    erie
    -1.75
    chel
    -1.55
    namespace
    -1.55
     hereto
    -1.48
    DEX
    -1.48
     hereinafter
    -1.48
    âķIJâķIJâķIJâķIJ
    -1.44
    ivo
    -1.44
    separated
    -1.43
     appropriate
    -1.42
    POSITIVE LOGITS
    ĻĤ
    2.45
    ĩ
    2.27
    ¼
    2.15
    ĥ
    2.12
    Ĥ¬
    2.08
     luck
    2.04
    ¾
    2.01
     enough
    1.97
    Ģ
    1.92
     habits
    1.89
    Activations Density 0.203%

    No Known Activations