© 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. 4-RES-JB
    11. 9089
    Prev
    Next
    INDEX
    Explanations

    mentions of law enforcement activities like raids

    oai_token-act-pair · gpt-3.5-turbo

    occurrences of the word "raided" and its variations in the context of police or government actions

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

    No Comments

    Negative Logits
    puff
    -0.73
    Perfect
    -0.71
    cong
    -0.70
    pure
    -0.68
    minecraft
    -0.67
     Supplement
    -0.64
    present
    -0.62
    chell
    -0.61
    ffen
    -0.61
    ankind
    -0.59
    POSITIVE LOGITS
     raided
    1.02
     raids
    1.00
     raid
    0.90
     sweeps
    0.87
     spree
    0.87
     raiding
    0.81
     seizure
    0.80
    nard
    0.78
     probes
    0.74
    itures
    0.73
    Activations Density 0.021%

    No Known Activations