© 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. 6838
    Prev
    Next
    INDEX
    Explanations

    URL paths, especially structured ones with repeated elements like "blah/blah/blah"

    oai_token-act-pair · gpt-3.5-turbo

    repeated sequences or patterns in text

    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
    Not in Any Lists

    No Comments

    Negative Logits
     pores
    -0.84
     creeping
    -0.76
     campaigning
    -0.73
     creep
    -0.72
     soaring
    -0.70
     hugging
    -0.70
     patrolling
    -0.70
     leaping
    -0.70
     piles
    -0.69
     eating
    -0.69
    POSITIVE LOGITS
    index
    1.48
    src
    1.20
    76561
    1.19
    comments
    1.18
    detail
    1.17
    Accessory
    1.15
    misc
    1.10
    modules
    1.10
    etc
    1.10
    addons
    1.09
    Activations Density 0.044%

    No Known Activations