© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI 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. 3-RES-JB
    11. 24126
    Prev
    Next
    INDEX
    Explanations

    sentences ending in periods

    oai_token-act-pair · gpt-3.5-turbo

    punctuation marks at the end of sentences

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 3-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    3
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    thal
    -0.76
     encount
    -0.73
    oshenko
    -0.73
    ãĥĭ
    -0.71
    ãĥ¼ãĥ
    -0.70
     purse
    -0.69
     exclusively
    -0.69
    onga
    -0.68
    zel
    -0.67
    ween
    -0.67
    POSITIVE LOGITS
     Especially
    1.13
     Luckily
    1.10
     Otherwise
    1.08
     Because
    1.08
     Obviously
    1.06
     However
    1.06
     Whether
    1.06
     But
    1.05
     Thankfully
    1.04
     If
    1.04
    Activations Density 1.111%

    No Known Activations