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. 14111
    Prev
    Next
    INDEX
    Explanations

    the word "scrap" or words related to discarding or reusing pieces of material

    oai_token-act-pair · gpt-3.5-turbo

    instances of the word "scrap" and its variations

    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
     Laws
    -0.70
    Mode
    -0.69
     Mode
    -0.66
    ception
    -0.65
     Modes
    -0.64
    lain
    -0.63
    ceive
    -0.61
     Metatron
    -0.61
    speaking
    -0.60
    orient
    -0.60
    POSITIVE LOGITS
     scrap
    3.98
     scraps
    2.33
     scra
    1.90
     scrape
    1.84
     scrapped
    1.74
     scraping
    1.51
     salvage
    1.39
     shred
    1.33
     salv
    1.29
     scratched
    1.16
    Activations Density 0.016%

    No Known Activations