© 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. 4810
    Prev
    Next
    INDEX
    Explanations

    information related to nutrient deficiencies

    oai_token-act-pair · gpt-3.5-turbo

    terms related to nutritional deficiencies and their effects

    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
    atom
    -0.78
    zees
    -0.71
    oller
    -0.68
    azon
    -0.67
    zee
    -0.67
    pper
    -0.67
    flo
    -0.66
    obo
    -0.66
    Oracle
    -0.65
    der
    -0.64
    POSITIVE LOGITS
     deficiency
    0.93
     deficiencies
    0.86
     defic
    0.78
     shortfall
    0.75
    iciency
    0.75
    icient
    0.75
     plag
    0.74
    xual
    0.72
    encies
    0.72
     Nigeria
    0.70
    Activations Density 0.012%

    No Known Activations