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

    Neuronpedia

    APIAssistant AxisNEWCircuit 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. 23531
    Prev
    Next
    INDEX
    Explanations

    word patterns with the suffix '-ansk' and 'Lanka'

    oai_token-act-pair · gpt-3.5-turbo

    geographical references, specifically places and names related to Ukraine and Sri Lanka

    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
     peer
    -0.68
     booth
    -0.67
     cour
    -0.62
    iler
    -0.59
    ipes
    -0.58
     printer
    -0.58
     cab
    -0.58
     weighing
    -0.57
     solic
    -0.57
     que
    -0.56
    POSITIVE LOGITS
    ansk
    4.66
    anka
    1.99
     Lanka
    1.53
    etsk
    1.43
    ANE
    1.16
    achev
    1.14
    inki
    1.07
     Gaal
    1.04
    oshenko
    1.02
    insk
    1.02
    Activations Density 0.008%

    No Known Activations