© 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. 22436
    Prev
    Next
    INDEX
    Explanations

    strings related to pharmaceuticals or health industries

    oai_token-act-pair · gpt-3.5-turbo

    instances of the acronym "AH" related to health organizations

    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
     Rend
    -0.66
     fing
    -0.65
     Stalin
    -0.64
     conj
    -0.63
     po
    -0.63
     Lac
    -0.62
    ktop
    -0.61
     quint
    -0.61
     determined
    -0.60
     diversity
    -0.60
    POSITIVE LOGITS
    AH
    4.42
     AH
    1.85
    ahs
    1.61
    ah
    1.57
    EH
    1.56
    OSH
    1.52
    aah
    1.50
    NH
    1.35
    AX
    1.34
    Ah
    1.32
    Activations Density 0.016%

    No Known Activations