© 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. 8015
    Prev
    Next
    INDEX
    Explanations

    A neuron 4 searches for the token "At" at the start of a sentence

    oai_token-act-pair · gpt-3.5-turbo

    phrases that start with "At" followed by a numerical value

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    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
     foul
    -0.64
     pudding
    -0.63
     deserved
    -0.61
     monopoly
    -0.60
     franch
    -0.60
     snap
    -0.58
     favor
    -0.58
     spoiled
    -0.58
     show
    -0.58
     divorce
    -0.58
    POSITIVE LOGITS
    At
    2.83
     At
    1.75
    During
    1.69
    As
    1.61
    After
    1.60
    Shortly
    1.57
    On
    1.51
    Of
    1.50
    From
    1.50
    With
    1.48
    Activations Density 0.018%

    No Known Activations