© 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. 2-RES-JB
    11. 1795
    Prev
    Next
    INDEX
    Explanations

    conjunctions signaling a contrast or contradiction

    oai_token-act-pair · gpt-3.5-turbo

    the conjunction "But" indicating contrasts or shifts in the narrative

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 2-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    2
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    heads
    -0.68
     segment
    -0.67
    broken
    -0.60
    interstitial
    -0.60
    abe
    -0.59
    ization
    -0.59
    ãĤĮ
    -0.59
    adr
    -0.58
    legged
    -0.58
    ãģ®
    -0.58
    POSITIVE LOGITS
    tons
    1.14
     alas
    0.94
     hey
    0.78
    chers
    0.75
    chery
    0.75
     luckily
    0.74
    ts
    0.71
    tery
    0.70
     nevertheless
    0.69
     fortunately
    0.67
    Activations Density 0.084%

    No Known Activations