© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW 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. 1-RES-JB
    11. 20411
    Prev
    Next
    INDEX
    Explanations

    quotes or reported speech

    oai_token-act-pair · gpt-3.5-turbo

    written dialogue, particularly sentences or phrases that denote speech

    oai_token-act-pair · gpt-4o-mini
    Triggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 1-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    1
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Negative Logits
     submerged
    -0.78
     purse
    -0.77
     metic
    -0.76
     totem
    -0.76
     halluc
    -0.73
     cob
    -0.73
     crab
    -0.72
     sublime
    -0.71
     holog
    -0.70
     unstoppable
    -0.68
    POSITIVE LOGITS
    ItemTracker
    0.96
    Downloadha
    0.90
     Afterwards
    0.90
     Then
    0.88
    Journal
    0.86
     But
    0.83
     Especially
    0.82
     Sources
    0.79
     Such
    0.77
    gov
    0.77
    Activations Density 1.167%

    No Known Activations