© 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. 12523
    Prev
    Next
    INDEX
    Explanations

    mentions of beverages like coffee, tea, champagne, and alcohol

    oai_token-act-pair · gpt-3.5-turbo

    quantities and references to items, particularly beverages and snacks

    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
    nw
    -0.80
    iets
    -0.76
    iland
    -0.72
    pmwiki
    -0.71
    ynes
    -0.70
    ahime
    -0.69
    abilities
    -0.68
    aucuses
    -0.68
    ername
    -0.67
    berman
    -0.66
    POSITIVE LOGITS
     sorts
    0.75
     cloth
    0.72
     maize
    0.71
     flats
    0.71
     glass
    0.71
     iron
    0.70
    icial
    0.69
     worms
    0.69
     orange
    0.69
     freshly
    0.67
    Activations Density 0.209%

    No Known Activations