© 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. 28579
    Prev
    Next
    INDEX
    Explanations

    structured data and code syntax

    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
    189
    +0.16
    0.9%
    214
    +0.12
    0.7%
    406
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    79
    +0.16
    0.10
    167
    +0.12
    0.09
    17
    +0.11
    0.02
    Negative Logits
    clock
    -1.48
    ally
    -1.43
     offices
    -1.41
    Fig
    -1.38
    fast
    -1.38
     bure
    -1.36
    coding
    -1.36
    awning
    -1.27
    amics
    -1.26
    alent
    -1.25
    POSITIVE LOGITS
    ĥ½
    2.10
    Ĺ
    1.74
    ĨĴ
    1.56
    į
    1.52
    ħ
    1.48
    İ
    1.47
    ij
    1.46
    ķ
    1.45
     retire
    1.45
    »
    1.44
    Activations Density 2.374%

    No Known Activations