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

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI 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. 4-RES-JB
    11. 10610
    Prev
    Next
    INDEX
    Explanations

    conjunctions followed by content that introduces a contrasting or opposite idea

    oai_token-act-pair · gpt-3.5-turbo

    instances of the word "But" indicating contrast or exception

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

    No Comments

    Negative Logits
    heads
    -0.69
     segment
    -0.68
    .","
    -0.61
    fell
    -0.60
     sym
    -0.59
    ¯¯¯¯
    -0.58
     ceremony
    -0.57
     award
    -0.56
    ization
    -0.56
     paths
    -0.55
    POSITIVE LOGITS
    tons
    1.33
    romeda
    0.93
     alas
    0.90
    theless
    0.88
    withstanding
    0.85
    thodox
    0.83
    chers
    0.83
    ts
    0.82
    tif
    0.79
    anamo
    0.78
    Activations Density 0.067%

    No Known Activations