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

    Neuronpedia

    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. 7-RES-JB
    11. 22964
    Prev
    Next
    INDEX
    Explanations

    phrases that end with 'ed' or ',' with high activation values

    oai_token-act-pair · gpt-3.5-turbo

    instances of commas followed by varying numerical or comparative expressions

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 7-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    7
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     Founding
    -0.72
    aha
    -0.70
    ¬¼
    -0.66
    ahu
    -0.65
    è£ħ
    -0.65
    ãģł
    -0.65
    inar
    -0.64
    asio
    -0.62
    ahs
    -0.62
    swick
    -0.61
    POSITIVE LOGITS
     albeit
    1.03
     whereas
    1.01
     except
    0.97
     although
    0.96
     however
    0.95
     but
    0.91
     owing
    0.91
     costing
    0.89
     according
    0.89
     prompting
    0.88
    Activations Density 0.287%

    No Known Activations