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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports 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. 16403
    Prev
    Next
    INDEX
    Explanations

    words related to intense negative emotions like being appalled, disgusted, or horrified

    oai_token-act-pair · gpt-3.5-turbo
    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
    -+
    -0.74
    Track
    -0.64
     main
    -0.63
     assets
    -0.62
     Status
    -0.62
    raft
    -0.61
    SS
    -0.61
    key
    -0.61
     commodity
    -0.61
    worth
    -0.61
    POSITIVE LOGITS
     appalled
    2.86
     horrified
    2.81
     disgusted
    2.62
     outraged
    2.56
     baffled
    2.54
     alarmed
    2.51
     angered
    2.44
     offended
    2.42
     astonished
    2.40
     puzzled
    2.32
    Activations Density 0.043%

    No Known Activations