Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Joseph Bloom · Open Source Sparse Autoencoders for all Residual Stream Layers of GPT2-Small
    3. GPT2-Small
    4. Residual Stream
    5. 0-RES-JB
    6. 3748
    Prev
    Next
    INDEX
    Explanations

    abbreviations and acronyms associated with news and media outlets, like "WS" for "Wall Street Journal"

    oai_token-act-pair · gpt-3.5-turbo

    mentions of "WS" (likely a reference to a news or sports organization) in the context of reporting or surveys

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 0-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    0
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     Leonard
    -0.67
     Lim
    -0.66
     lime
    -0.63
     villain
    -0.62
     cast
    -0.62
     exc
    -0.61
     comple
    -0.60
     Finland
    -0.60
     lev
    -0.59
     lim
    -0.59
    POSITIVE LOGITS
    WS
    4.54
     WS
    2.11
    ws
    1.85
    WC
    1.73
    WN
    1.55
    WP
    1.54
    WB
    1.48
    WF
    1.42
    Ws
    1.42
    WD
    1.39
    Activations Density 0.014%

    No Known Activations