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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & 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. 17897
    Prev
    Next
    INDEX
    Explanations

    instances where something is done in a different way than expected or where an exception applies

    oai_token-act-pair · gpt-3.5-turbo

    phrases that introduce exceptions or alternatives

    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
    aire
    -0.66
    ression
    -0.65
    king
    -0.59
     Bunker
    -0.56
    mare
    -0.55
     Hoo
    -0.54
    edes
    -0.53
     Among
    -0.53
    shock
    -0.52
    Quotes
    -0.51
    POSITIVE LOGITS
     otherwise
    4.28
     Otherwise
    2.11
    Otherwise
    1.89
     ordinarily
    1.46
     normally
    1.30
     else
    1.22
     hitherto
    1.10
     elsewhere
    1.02
    Else
    1.01
     Else
    0.99
    Activations Density 0.019%

    No Known Activations