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. Apollo Research · Taylor · Identifying Functionally Important Features with End-to-End Sparse Dictionary Learning
    3. GPT2-Small
    4. Residual Similar CE Local
    5. 6-RES_SCL-AJT
    6. 45210
    Prev
    Next
    INDEX
    Explanations

    proper nouns and titles, particularly related to political organizations and events

    oai_token-act-pair · gpt-3.5-turbo
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    neuronpedia/gpt2-small__res_scl-ajt/6-res_scl-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
    447
    +0.11
    0.6%
    665
    +0.10
    0.5%
    479
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    479
    +0.11
    0.04
    288
    +0.10
    0.24
    2
    +0.10
    0.05
    Negative Logits
    angible
    -0.80
    Ô
    -0.79
    bda
    -0.71
    nos
    -0.71
    ho
    -0.69
     sim
    -0.68
    make
    -0.67
    roc
    -0.62
    sim
    -0.61
    tele
    -0.61
    POSITIVE LOGITS
     Agency
    0.89
     Desk
    0.85
     Relief
    0.79
     Protection
    0.79
     Inquisition
    0.79
     Registry
    0.79
     Revival
    0.79
     Aid
    0.78
     Patrol
    0.77
     Initiative
    0.77
    Activations Density 11.708%

    No Known Activations