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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & 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. 28936
    Prev
    Next
    INDEX
    Explanations

    phrases indicating free software and its licensing terms

    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
    74
    +0.15
    0.8%
    107
    +0.14
    0.8%
    289
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    74
    +0.15
    0.02
    107
    +0.14
    0.02
    289
    +0.10
    0.02
    Negative Logits
    ĸ
    -3.06
    ľĵ
    -3.00
    ĨĴ
    -2.92
    ¢
    -2.79
    ı
    -2.76
    Ĺ
    -2.72
    Ĥ
    -2.70
                                                                                  
    -2.69
                             
    -2.69
     ↵       
    -2.69
    POSITIVE LOGITS
     reagent
    1.64
     axi
    1.51
     confidential
    1.46
     Elementary
    1.45
     specialty
    1.45
     expire
    1.40
     adj
    1.38
     expired
    1.37
     abelian
    1.37
     inactivated
    1.33
    Activations Density 0.078%

    No Known Activations