© 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. Joseph Bloom · Open Source Sparse Autoencoders for all Residual Stream Layers of GPT2-Small
    5. GPT2-Small
    7. Residual Stream
    9. 0-RES-JB
    11. 15230
    Prev
    Next
    INDEX
    Explanations

    the word "appealed" or similar words indicating legal recourse or challenge

    oai_token-act-pair · gpt-3.5-turbo

    instances of the word "appealed."

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 0-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    0
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     shred
    -0.77
     chop
    -0.72
     select
    -0.67
     exit
    -0.66
     spin
    -0.64
     hij
    -0.63
    asure
    -0.63
    ning
    -0.62
     nap
    -0.60
     Pierce
    -0.60
    POSITIVE LOGITS
     appealed
    3.73
     clicked
    1.59
     intervened
    1.59
     vetoed
    1.44
     objected
    1.38
     defied
    1.36
     communicated
    1.35
     interfered
    1.32
     protested
    1.31
     affirmed
    1.26
    Activations Density 0.016%

    No Known Activations