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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & 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. 34090
    Prev
    Next
    INDEX
    Explanations

    mentions of chemical substances, supplements, and pharmaceuticals

    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
    642
    +0.06
    0.5%
    618
    +0.06
    0.5%
    554
    +0.06
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    288
    +0.06
    0.23
    326
    +0.06
    0.20
    549
    +0.06
    0.08
    Negative Logits
    emb
    -0.41
    ille
    -0.39
    akia
    -0.39
    cats
    -0.38
    ocene
    -0.38
    ĸ
    -0.38
    acht
    -0.38
    boat
    -0.38
    ILLE
    -0.37
    76561
    -0.37
    POSITIVE LOGITS
    aceutical
    0.42
    undown
    0.41
    atively
    0.40
    ividual
    0.40
    iously
    0.39
    acebook
    0.39
     augment
    0.38
    mented
    0.38
    atives
    0.38
     athletics
    0.38
    Activations Density 7.023%

    No Known Activations