© 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. 0-RES-JB
    11. 16389
    Prev
    Next
    INDEX
    Explanations

    phrases related to events or actions happening before a specific point in time

    oai_token-act-pair · gpt-3.5-turbo
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 0-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    0
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     twisting
    -0.68
     humour
    -0.68
     charming
    -0.68
     snaps
    -0.68
     drag
    -0.67
     laugh
    -0.65
     beautiful
    -0.65
     trick
    -0.65
     adorable
    -0.65
     comedy
    -0.64
    POSITIVE LOGITS
     Prior
    3.68
    Prior
    2.82
     Previously
    1.56
     prior
    1.43
     Previous
    1.30
     Before
    1.23
    IOR
    1.13
     Earlier
    1.05
     Priority
    1.04
    Previous
    1.03
    Activations Density 0.020%

    No Known Activations