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. 23598
    Prev
    Next
    INDEX
    Explanations

    Russian Cyrillic characters and possibly some English phrases mixed with Cyrillic characters, but there doesn't seem to be a clear pattern or specific content it's looking for in the available context

    oai_token-act-pair · gpt-3.5-turbo

    high-frequency or repeated characters and patterns in non-Latin scripts

    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
     pac
    -0.76
     gall
    -0.73
     Directions
    -0.72
     GT
    -0.67
     panel
    -0.66
     Indiana
    -0.65
     wand
    -0.64
     Wa
    -0.64
     Chal
    -0.64
     TG
    -0.63
    POSITIVE LOGITS
    е
    4.21
    а
    2.80
    о
    2.75
    и
    2.57
    оÐ
    2.50
    Ñĥ
    2.46
    Ñ
    2.38
    Ñı
    2.34
    Ñĭ
    2.32
    ÑĮ
    2.20
    Activations Density 0.004%

    No Known Activations