© 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. 18279
    Prev
    Next
    INDEX
    Explanations

    time-related words or phrases, particularly those related to counting seconds

    oai_token-act-pair · gpt-3.5-turbo

    instances of the word "seconds" associated with various contexts or durations

    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
    rim
    -0.70
    oret
    -0.69
    hemoth
    -0.67
    ramid
    -0.65
    ebook
    -0.64
    hammer
    -0.64
    iHUD
    -0.63
    duct
    -0.62
    hus
    -0.62
    ãĥĢ
    -0.60
    POSITIVE LOGITS
     elapsed
    1.15
    aneously
    0.92
     ago
    0.88
     seconds
    0.87
     msec
    0.86
    seconds
    0.84
    ilver
    0.80
     Seconds
    0.80
    onds
    0.79
     mins
    0.76
    Activations Density 0.018%

    No Known Activations