© 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. 8-RES-JB
    11. 1890
    Prev
    Next
    INDEX
    Explanations

    instances where two entities, actions, or objects are being compared or contrasted

    oai_token-act-pair · gpt-3.5-turbo

    instances where the word "one" is used to denote individual items in contrast to "two" or others

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    8
    Activation Function
    relu
    Embeds
    IFrame
    Link
    •all features

    No Comments

    Negative Logits
    uits
    -0.82
    uit
    -0.80
    renches
    -0.77
    hip
    -0.76
    models
    -0.75
    hips
    -0.75
    ilater
    -0.73
    acements
    -0.73
    iculty
    -0.70
    stru
    -0.68
    POSITIVE LOGITS
     apiece
    0.99
    ERA
    0.67
     thing
    0.67
     detects
    0.66
     assumes
    0.66
     side
    0.63
    IDA
    0.63
     accuser
    0.62
     lobe
    0.62
     exception
    0.61
    Activations Density 0.106%

    No Known Activations