© 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. 6-RES-JB
    11. 14977
    Prev
    Next
    INDEX
    Explanations

    terms related to hierarchical ranking and organization

    oai_token-act-pair · gpt-3.5-turbo

    phrases indicating ranking or hierarchy

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 6-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.6.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.6.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    6
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    enegger
    -0.76
    juven
    -0.72
    reddits
    -0.70
    interstitial
    -0.69
    gest
    -0.68
    netflix
    -0.68
    :[
    -0.67
    voy
    -0.67
    >[
    -0.67
    iors
    -0.66
    POSITIVE LOGITS
     parcel
    0.75
    rogens
    0.73
     therefore
    0.72
    rogen
    0.70
    romeda
    0.68
     consequently
    0.66
     pound
    0.64
     then
    0.63
     Conquer
    0.62
     mortar
    0.61
    Activations Density 0.180%

    No Known Activations