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. 3425
    Prev
    Next
    INDEX
    Explanations

    words related to a specific concept or topic, potentially related to legal, scholarly, or technical subjects

    oai_token-act-pair · gpt-3.5-turbo

    the term "ament" and its variations, indicating a focus on concepts related to causes, consequences, or structures that are fundamental or foundational

    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
     sher
    -0.73
     Pri
    -0.66
    Ry
    -0.61
     rs
    -0.61
     reporter
    -0.60
     Rogue
    -0.60
     Twins
    -0.59
     Warden
    -0.58
     jog
    -0.58
     Ry
    -0.58
    POSITIVE LOGITS
    ament
    4.91
    aments
    3.80
    amental
    2.61
    ement
    1.19
    ments
    1.17
    iment
    1.10
    ements
    1.08
    ment
    1.03
    ature
    1.01
    ument
    0.99
    Activations Density 0.015%

    No Known Activations