© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Apollo Research · Taylor · Identifying Functionally Important Features with End-to-End Sparse Dictionary Learning
    5. GPT2-Small
    7. Residual Similar CE E2E Recon
    9. 6-RES_SCEFR-AJT
    11. 13427
    Prev
    Next
    INDEX
    Explanations

    opposite pairs of concepts like good and evil, light and darkness, sweet and bitter, which are often associated with strong emotional or moral contrasts

    oai_token-act-pair · gpt-3.5-turbo
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    neuronpedia/gpt2-small__res_scefr-ajt/6-res_scefr-ajt
    Prompts (Dashboard)
    12,288 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    46,080
    Data Type
    torch.float32
    Hook Point
    blocks.6.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    apollo-research/Skylion007-openwebtext-tokenizer-gpt2
    Hook Point Layer
    6
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    266
    +0.54
    7.5%
    480
    +0.17
    2.4%
    326
    +0.08
    1.1%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    480
    +0.54
    -0.30
    266
    +0.17
    0.94
    351
    +0.08
    0.03
    Negative Logits
    soDeliveryDate
    -0.29
    BuyableInstoreAndOnline
    -0.24
    âĹ¼
    -0.23
     horm
    -0.21
    hov
    -0.21
    ongo
    -0.21
    oola
    -0.20
    ãĥ´
    -0.20
    uberty
    -0.20
     Gleaming
    -0.20
    POSITIVE LOGITS
     yours
    0.19
     publicly
    0.19
     legally
    0.18
     completely
    0.17
     liter
    0.17
     entirely
    0.17
     hers
    0.17
     unused
    0.17
     dangerously
    0.17
     ours
    0.17
    Activations Density 98.227%

    No Known Activations