© 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. 15292
    Prev
    Next
    INDEX
    Explanations

    the word "az" with varying activation levels

    oai_token-act-pair · gpt-3.5-turbo

    references to a specific person named "Az."

    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
    nces
    -0.70
     Fargo
    -0.67
     Kraken
    -0.65
     foremost
    -0.65
     Kepler
    -0.65
     Norn
    -0.63
    ACTED
    -0.62
     Impossible
    -0.61
     Kardashian
    -0.61
    waters
    -0.60
    POSITIVE LOGITS
    ombie
    1.15
    hou
    1.04
    hang
    0.99
    quez
    0.98
    eez
    0.97
    illion
    0.96
    azel
    0.96
    awa
    0.95
    quet
    0.92
    az
    0.91
    Activations Density 0.016%

    No Known Activations