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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & 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. 8-RES-JB
    6. 22721
    Prev
    Next
    INDEX
    Explanations

    mentions of legal entities and court-related terms

    oai_token-act-pair · gpt-3.5-turbo

    references to legal jurisdictions and federal court cases

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    8
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     sacrific
    -0.86
     boosts
    -0.78
     flashes
    -0.78
     wors
    -0.77
    Reviewer
    -0.77
    liness
    -0.75
    dyl
    -0.72
    reddits
    -0.72
     indo
    -0.70
     sleeps
    -0.70
    POSITIVE LOGITS
    Attorney
    1.18
    Court
    1.16
     court
    1.12
     indictment
    1.11
     attorney
    1.09
     Attorney
    1.09
     extradition
    1.09
     Court
    1.07
     Judicial
    1.05
     attorneys
    1.05
    Activations Density 0.323%

    No Known Activations