© 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. 20169
    Prev
    Next
    INDEX
    Explanations

    instances where the phrase "could not" is used in text

    oai_token-act-pair · gpt-3.5-turbo

    phrases that indicate an inability or failure to do something

    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
    éĥ
    -0.69
    Intern
    -0.66
    ership
    -0.66
    SI
    -0.66
    SER
    -0.65
    AFP
    -0.64
    pages
    -0.64
    anders
    -0.64
    Maker
    -0.64
    å¥
    -0.63
    POSITIVE LOGITS
    icably
    1.38
     necessarily
    1.33
    icable
    1.17
     bother
    0.99
     hesitate
    0.96
    epad
    0.95
     tolerate
    0.90
     bud
    0.86
    necess
    0.85
     exactly
    0.84
    Activations Density 0.211%

    No Known Activations