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. 14396
    Prev
    Next
    INDEX
    Explanations

    programming code with specific patterns and syntax

    oai_token-act-pair · gpt-3.5-turbo

    phrases and syntax commonly used in programming or technical contexts, particularly the arrow function notation in JavaScript

    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
     Paran
    -0.61
     veteran
    -0.61
     Sty
    -0.61
     Hin
    -0.59
     Vernon
    -0.59
     Nich
    -0.58
     sle
    -0.57
    reck
    -0.57
     delusions
    -0.57
     Piet
    -0.57
    POSITIVE LOGITS
     =>
    4.28
    =>
    2.61
     ->
    1.54
     ===
    1.36
     âĨĴ
    1.29
     âĩ
    1.20
    "]=>
    1.16
     -->
    1.12
     âī¡
    1.11
     âĶĤ
    1.07
    Activations Density 0.007%

    No Known Activations