© 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. 3-RES-JB
    11. 12514
    Prev
    Next
    INDEX
    Explanations

    words related to policies, regulations, and government activities

    oai_token-act-pair · gpt-3.5-turbo

    phrases related to environmental and cultural themes

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 3-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    3
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    ãĥ¼ãĥĨãĤ£
    -0.45
    ãĥ¼ãĥĨ
    -0.44
    ãĥ¼ãĥ³
    -0.43
    ogether
    -0.41
    ãĥĥãĥī
    -0.39
    aturdays
    -0.38
     referen
    -0.37
    emale
    -0.37
    $$$$
    -0.37
     challeng
    -0.36
    POSITIVE LOGITS
    NetMessage
    0.37
    rg
    0.37
     DRAGON
    0.37
    bleacher
    0.36
     largeDownload
    0.36
    CVE
    0.36
    esson
    0.34
    terness
    0.34
     REUTERS
    0.33
    ktop
    0.32
    Activations Density 3.954%

    No Known Activations