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

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & 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. 11424
    Prev
    Next
    INDEX
    Explanations

    references to authors and copyright information

    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
    271
    +0.13
    0.8%
    174
    +0.12
    0.7%
    168
    +0.11
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    168
    +0.13
    0.02
    222
    +0.12
    0.02
    298
    +0.11
    0.02
    Negative Logits
    MRI
    -1.44
    REE
    -1.42
    correct
    -1.41
    electric
    -1.37
    bil
    -1.36
    true
    -1.35
    odor
    -1.35
     typical
    -1.35
     yes
    -1.34
    cele
    -1.33
    POSITIVE LOGITS
    eman
    1.44
    аеÑĤ
    1.42
     constants
    1.41
    ãĥ¥
    1.39
    imes
    1.38
    аÑĤел
    1.38
     Enterprise
    1.38
    Ñĭ
    1.37
    hu
    1.36
    imus
    1.36
    Activations Density 0.111%

    No Known Activations