© 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. 0-RES-JB
    11. 9971
    Prev
    Next
    INDEX
    Explanations

    references to directions or instructions given by someone

    oai_token-act-pair · gpt-3.5-turbo

    instances of the word "directed" 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
    itol
    -0.72
    odox
    -0.69
     coff
    -0.66
     Register
    -0.65
    maxwell
    -0.65
    breaker
    -0.65
    yrs
    -0.63
    Numbers
    -0.62
    pick
    -0.62
     Breaker
    -0.61
    POSITIVE LOGITS
     directed
    3.70
    directed
    2.27
     directs
    1.98
     directing
    1.95
     redirected
    1.58
     instructed
    1.42
     aimed
    1.41
     commanded
    1.37
    Direct
    1.37
     ordered
    1.25
    Activations Density 0.016%

    No Known Activations