© 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. 2-RES-JB
    11. 24208
    Prev
    Next
    INDEX
    Explanations

    number sequences in the format xx that are repeated frequently

    oai_token-act-pair · gpt-3.5-turbo

    occurrences of timestamps or numerical representations

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 2-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    2
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    inventoryQuantity
    -0.82
    ascus
    -0.74
    loo
    -0.74
    izont
    -0.71
     offended
    -0.71
    matically
    -0.70
    DragonMagazine
    -0.69
    fired
    -0.67
    kell
    -0.67
    Kin
    -0.67
    POSITIVE LOGITS
     00
    1.33
     01
    0.97
    een
    0.96
    eenth
    0.85
     03
    0.84
     02
    0.84
    ees
    0.81
     04
    0.80
     05
    0.79
     000
    0.78
    Activations Density 0.008%

    No Known Activations