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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports 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. 10967
    Prev
    Next
    INDEX
    Explanations

    phrases related to contrasting information or exceptions

    oai_token-act-pair · gpt-3.5-turbo

    phrases that begin with "however," indicating a 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
    chairs
    -0.67
    roy
    -0.67
    MJ
    -0.62
    om
    -0.59
    OW
    -0.59
    ore
    -0.59
    ULAR
    -0.58
    Cover
    -0.56
    kay
    -0.56
     nightmare
    -0.54
    POSITIVE LOGITS
     acknow
    0.77
     //[
    0.76
     suffice
    0.74
    âķIJâķIJ
    0.71
     alas
    0.70
     namely
    0.68
    DragonMagazine
    0.66
    acha
    0.66
    ilation
    0.66
    aido
    0.66
    Activations Density 0.103%

    No Known Activations