© 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 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. 0-RES-JB
    11. 12547
    Prev
    Next
    INDEX
    Explanations

    researchers or research-related terms

    oai_token-act-pair · gpt-3.5-turbo

    mentions of researchers and their studies

    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
     representation
    -0.76
     Creed
    -0.76
     transition
    -0.76
     loyalty
    -0.72
     uniform
    -0.70
     duty
    -0.68
     repr
    -0.68
     Franchise
    -0.67
     JR
    -0.65
     Fal
    -0.65
    POSITIVE LOGITS
    Researchers
    2.96
    Scientists
    2.17
    Ingredients
    1.69
    Experts
    1.60
    Research
    1.60
    rawdownloadcloneembedreportprint
    1.55
     Researchers
    1.55
    Engineers
    1.54
    Study
    1.45
    Redditor
    1.44
    Activations Density 0.045%

    No Known Activations