© 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. 8-RES-JB
    11. 10004
    Prev
    Next
    INDEX
    Explanations

    phrases related to gathering information or pieces and putting them together into a cohesive whole

    oai_token-act-pair · gpt-3.5-turbo

    phrases related to processing and organizing information

    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
    guard
    -0.91
     Passage
    -0.79
    osponsors
    -0.79
    guards
    -0.79
    airs
    -0.75
    taker
    -0.71
    writers
    -0.71
    Guard
    -0.70
    nor
    -0.70
    reb
    -0.65
    POSITIVE LOGITS
     disparate
    1.15
     together
    1.02
     fragments
    0.96
     leftover
    0.94
     ashes
    0.91
     rubble
    0.90
     scraps
    0.88
     fragmented
    0.88
     data
    0.85
     rubbish
    0.84
    Activations Density 0.605%

    No Known Activations