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

    Neuronpedia

    APIAssistant AxisNEWCircuit 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. 12010
    Prev
    Next
    INDEX
    Explanations

    specific legal terms and concepts related to criminal offenses

    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
    443
    +0.13
    0.7%
    161
    +0.13
    0.7%
    261
    +0.12
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    443
    +0.13
    0.08
    74
    +0.13
    0.07
    391
    +0.12
    0.07
    Negative Logits
     naming
    -1.59
    shell
    -1.40
     preparation
    -1.36
     granted
    -1.35
    osis
    -1.35
    etting
    -1.34
    otomy
    -1.31
     normalization
    -1.31
     preparations
    -1.30
    osing
    -1.27
    POSITIVE LOGITS
    ¬
    4.16
    Ļª
    3.99
    ³
    3.98
    ¦
    3.96
    ±
    3.90
    ĥ½
    3.85
    ª
    3.82
    ĸ
    3.77
    ŀ
    3.76
    §
    3.75
    Activations Density 2.967%

    No Known Activations