© 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. 21093
    Prev
    Next
    INDEX
    Explanations

    legal language related to evidence and justification

    oai_token-act-pair · gpt-3.5-turbo

    terms associated with legal justification and evidentiary standards

    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
    erie
    -0.93
    ollen
    -0.77
    eteria
    -0.77
    utch
    -0.76
    reb
    -0.76
    ffield
    -0.75
    osi
    -0.74
    ework
    -0.73
    assault
    -0.72
    perors
    -0.72
    POSITIVE LOGITS
     warrant
    0.87
     cause
    0.83
     warrants
    0.82
     probable
    0.82
     disposition
    0.80
     indication
    0.77
    ãĥ¼ãĥĨ
    0.72
     affidavit
    0.71
     suspect
    0.69
     assumption
    0.68
    Activations Density 0.008%

    No Known Activations