Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Joseph Bloom · Open Source Sparse Autoencoders for all Residual Stream Layers of GPT2-Small
    3. GPT2-Small
    4. Residual Stream
    5. 0-RES-JB
    6. 3141
    Prev
    Next
    INDEX
    Explanations

    words related to legal and criminal matters, specifically focusing on terms related to criminal offenders and sentencing

    oai_token-act-pair · gpt-3.5-turbo

    references to individuals who have committed crimes or offenses

    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
    ma
    -0.69
    Mom
    -0.66
    ype
    -0.61
    jet
    -0.61
    ira
    -0.60
    tz
    -0.58
    ruce
    -0.57
    fig
    -0.57
     Walt
    -0.56
     Sur
    -0.56
    POSITIVE LOGITS
     offenders
    3.72
     offender
    3.55
     abusers
    1.84
     perpetrators
    1.83
     offending
    1.76
     culprit
    1.70
     offenses
    1.64
     criminals
    1.63
     offences
    1.63
     perpetrator
    1.54
    Activations Density 0.013%

    No Known Activations