© 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. 10423
    Prev
    Next
    INDEX
    Explanations

    connections or associations between entities or concepts

    oai_token-act-pair · gpt-3.5-turbo

    phrases indicating connections or relationships between people or entities

    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
    oche
    -0.77
    NEY
    -0.73
    AH
    -0.73
    INESS
    -0.71
    needed
    -0.70
    Ĥİ
    -0.70
    iana
    -0.69
    escription
    -0.68
    jing
    -0.68
    SHA
    -0.68
    POSITIVE LOGITS
     halves
    0.85
     thence
    0.77
     extremes
    0.71
     grasp
    0.69
     vice
    0.66
    ingo
    0.65
     aspirations
    0.64
     Ferguson
    0.64
     vantage
    0.62
     practise
    0.62
    Activations Density 0.109%

    No Known Activations