© 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 EvalsExportsAPI 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. 10537
    Prev
    Next
    INDEX
    Explanations

    terms related to physical appearances and attributes

    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
    153
    +0.18
    1.0%
    326
    +0.13
    0.7%
    385
    +0.12
    0.7%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    153
    +0.18
    0.07
    413
    +0.13
    0.03
    56
    +0.12
    -0.03
    Negative Logits
     forums
    -1.80
     forum
    -1.80
     program
    -1.76
    journal
    -1.76
     policy
    -1.71
    ournal
    -1.61
    forum
    -1.61
    ttp
    -1.55
    iner
    -1.53
     talks
    -1.53
    POSITIVE LOGITS
     stained
    2.20
    »¿
    1.96
     carc
    1.89
     flank
    1.87
     stain
    1.85
     elongated
    1.83
     ornament
    1.80
     lobes
    1.78
    washed
    1.75
     stains
    1.74
    Activations Density 2.239%

    No Known Activations