© 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. 8-RES-JB
    11. 6231
    Prev
    Next
    INDEX
    Explanations

    the word "do" in sentences

    oai_token-act-pair · gpt-3.5-turbo

    repetitive phrases indicating agreement or affirmation

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    8
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     Madagascar
    -0.69
    case
    -0.65
    ufact
    -0.64
     Entered
    -0.62
    hog
    -0.62
     SAR
    -0.62
     Printed
    -0.62
    aro
    -0.61
     Handling
    -0.60
     Hok
    -0.59
    POSITIVE LOGITS
    vet
    0.92
    pez
    0.89
     apologise
    0.86
     understand
    0.85
    ppel
    0.84
     intend
    0.83
    berman
    0.83
     acknowledge
    0.82
     apologize
    0.82
     hereby
    0.81
    Activations Density 0.092%

    No Known Activations