© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Joseph Bloom · Open Source Sparse Autoencoders for all Residual Stream Layers of GPT2-Small
    5. GPT2-Small
    7. Residual Stream
    9. 3-RES-JB
    11. 5002
    Prev
    Next
    INDEX
    Explanations

    adjectives related to qualities such as being wrong, required, wrong, or concerned

    oai_token-act-pair · gpt-3.5-turbo

    terms related to societal challenges and responsibilities

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 3-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    3
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     VIDEOS
    -0.68
    Previous
    -0.67
    ourage
    -0.66
    ederal
    -0.63
    ajo
    -0.62
    ario
    -0.62
    dress
    -0.61
    éĹ
    -0.60
    achu
    -0.59
    osion
    -0.59
    POSITIVE LOGITS
    abouts
    0.91
     interchangeable
    0.80
     li
    0.75
     themselves
    0.73
     diabetic
    0.72
    omorphic
    0.71
    entary
    0.67
    cerned
    0.65
     jelly
    0.64
    ometers
    0.64
    Activations Density 0.983%

    No Known Activations