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

    Neuronpedia

    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. 0-RES-JB
    11. 18904
    Prev
    Next
    INDEX
    Explanations

    negative words and phrases

    oai_token-act-pair · gpt-3.5-turbo

    occurrences of the word "bad" used in various contexts

    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
    icum
    -0.82
    adel
    -0.81
    rise
    -0.77
    fold
    -0.75
    endant
    -0.73
    ettes
    -0.72
    aternity
    -0.70
    itudinal
    -0.69
    cry
    -0.67
    kamp
    -0.66
    POSITIVE LOGITS
     bad
    3.34
    bad
    2.39
     BAD
    2.18
     Bad
    2.05
    Bad
    2.03
     terrible
    1.85
     lousy
    1.73
     crappy
    1.66
     horrible
    1.65
     shitty
    1.58
    Activations Density 0.019%

    No Known Activations