© 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
    2. Joseph Bloom · Open Source Sparse Autoencoders for all Residual Stream Layers of GPT2-Small
    3. GPT2-Small
    4. Residual Stream
    5. 0-RES-JB
    6. 11773
    Prev
    Next
    INDEX
    Explanations

    instances of the word "friendly" with a strong match

    oai_token-act-pair · gpt-3.5-turbo

    the adjective "friendly" in various contexts

    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
    pieces
    -0.71
    pta
    -0.68
    conom
    -0.66
    mith
    -0.65
    tor
    -0.65
    Auth
    -0.64
    court
    -0.63
    ulla
    -0.63
    ham
    -0.63
    edi
    -0.62
    POSITIVE LOGITS
     friendly
    3.88
     Friendly
    2.90
    friendly
    2.51
    Friend
    1.91
     hostile
    1.71
     cheerful
    1.28
     friend
    1.27
     respectful
    1.27
    riend
    1.27
     polite
    1.25
    Activations Density 0.021%

    No Known Activations