© 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. 3-RES-JB
    11. 23415
    Prev
    Next
    INDEX
    Explanations

    contractions combined with verbs, such as "don't," "didn't," "wouldn't."

    oai_token-act-pair · gpt-3.5-turbo

    negative contractions and phrases indicating failure or refusal

    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
    heed
    -0.63
    currently
    -0.61
    now
    -0.61
    arta
    -0.60
    onto
    -0.60
    ologies
    -0.60
    lies
    -0.60
    bits
    -0.58
    Member
    -0.58
    requires
    -0.57
    POSITIVE LOGITS
     last
    0.79
     originally
    0.69
     earlier
    0.69
     yesterday
    0.69
     awaken
    0.60
    iscons
    0.59
     booted
    0.59
     initially
    0.58
     recently
    0.57
     unsuccessfully
    0.56
    Activations Density 2.365%

    No Known Activations