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. 8-RES-JB
    6. 3794
    Prev
    Next
    INDEX
    Explanations

    compound words

    oai_token-act-pair · gpt-3.5-turbo

    terms related to events and productions in entertainment

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    8
    Activation Function
    relu
    Embeds
    IFrame
    Link
    •contextual features
    Fires on multi-token proper nouns (lots of contexts including video games, TV, music, anime). Fires on "DT", " Odyssey", " HUD", " Battlefield", "Action"

    No Comments

    Negative Logits
     destro
    -0.70
     helicop
    -0.69
     advoc
    -0.66
     withd
    -0.66
    essage
    -0.65
     arrang
    -0.65
     undermin
    -0.65
    Ò
    -0.63
    hovah
    -0.63
    ̶
    -0.62
    POSITIVE LOGITS
     Profile
    1.16
     âĵĺ
    0.92
     Brewing
    0.82
     Overview
    0.81
     Releases
    0.81
    ¶
    0.79
     Forums
    0.77
     Calculator
    0.77
     Facts
    0.75
     Finder
    0.74
    Activations Density 0.569%

    No Known Activations