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

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI 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. 7838
    Prev
    Next
    INDEX
    Explanations

    discussions related to software development, debugging, and coding errors

    oai_token-act-pair · gpt-3.5-turbo

    punctuation marks such as parentheses, periods, and commas, indicating it may be focused on sentence structure and punctuation

    oai_token-act-pair · gpt-3.5-turboTriggered by @jordan
    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
    640
    +0.20
    0.9%
    679
    +0.17
    0.8%
    393
    +0.14
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    679
    +0.20
    0.25
    6
    +0.17
    0.23
    640
    +0.14
    0.23
    Negative Logits
     monop
    -0.86
     aven
    -0.81
     sands
    -0.78
    lies
    -0.77
     advoc
    -0.76
    jug
    -0.76
    ozo
    -0.75
    natureconservancy
    -0.73
     thous
    -0.73
     gobl
    -0.71
    POSITIVE LOGITS
     Conversely
    1.17
     Whereas
    1.17
     Often
    1.10
     Likewise
    1.08
     Usually
    1.08
     Especially
    1.06
     Similarly
    1.04
     Different
    1.04
     Alternatively
    1.00
     Depending
    0.99
    Activations Density 2.650%

    No Known Activations