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. 20620
    Prev
    Next
    INDEX
    Explanations

    instances where the phrases "to have," "to keep," or "to pack" are mentioned alongside specific objects or actions

    oai_token-act-pair · gpt-3.5-turbo

    references to violence or sexual themes

    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
     largeDownload
    -0.85
    natureconservancy
    -0.84
    etheless
    -0.78
    amination
    -0.69
    notations
    -0.68
    enaries
    -0.67
    urgy
    -0.67
    ignty
    -0.67
    ŃĶ
    -0.65
     Pwr
    -0.65
    POSITIVE LOGITS
    OHN
    0.72
    oston
    0.71
     pestic
    0.64
    pload
    0.64
    Nitrome
    0.62
    DERR
    0.61
    irlf
    0.59
     Niagara
    0.59
     Augusta
    0.58
    VALUE
    0.57
    Activations Density 6.243%

    No Known Activations