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. 45317
    Prev
    Next
    INDEX
    Explanations

    mentions of specific names, likely related to legal matters or news stories

    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
    •Layer 6 UMAP region: Mostly-local cluster on left - local

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    447
    +0.20
    0.8%
    649
    +0.14
    0.6%
    163
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    146
    +0.20
    0.09
    521
    +0.14
    0.02
    639
    +0.11
    0.08
    Negative Logits
    SPONSORED
    -1.00
    mble
    -0.86
    qqa
    -0.83
    igrate
    -0.77
    orsi
    -0.74
    elvet
    -0.72
    raint
    -0.70
    OPE
    -0.70
    uality
    -0.68
    idad
    -0.67
    POSITIVE LOGITS
    products
    1.11
    product
    0.97
    laws
    0.92
    gone
    0.85
    dos
    0.84
    law
    0.77
    pass
    0.75
    akuya
    0.75
    stant
    0.75
    Products
    0.75
    Activations Density 9.071%

    No Known Activations