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

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Joseph Bloom · Open Source Sparse Autoencoders for all Residual Stream Layers of GPT2-Small
    3. GPT2-Small
    4. Residual Stream
    5. 1-RES-JB
    6. 13672
    Prev
    Next
    INDEX
    Explanations

    I'm sorry, but I cannot provide a summary for Neuron 4's behavior as there seems to be a variety of strings and words with the common substring "Ord". Would you like me to attempt a different approach to identify the pattern characteristic for this neuron?

    oai_token-act-pair · gpt-3.5-turbo

    words related to ordinations and religious titles

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 1-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.hook_resid_pre
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    24,576
    Data Type
    torch.float32
    Hook Point
    blocks.1.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    1
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    vous
    -0.75
    ï¸ı
    -0.74
    =-=-=-=-
    -0.70
     Libre
    -0.69
    ãĥĨãĤ£
    -0.67
    STEM
    -0.67
    PsyNetMessage
    -0.67
    GROUND
    -0.65
    icago
    -0.65
    Explore
    -0.64
    POSITIVE LOGITS
    nance
    1.51
    inarily
    1.39
    inances
    1.27
    inary
    1.19
    inance
    1.03
     Ord
    0.99
    inal
    0.98
    eals
    0.92
    ered
    0.91
     ord
    0.91
    Activations Density 0.003%

    No Known Activations