© 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. 3-RES-JB
    11. 8347
    Prev
    Next
    INDEX
    Explanations

    phrases related to legal actions or court proceedings

    oai_token-act-pair · gpt-3.5-turbo

    instances of the word "try" and its variations related to legal proceedings

    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
    oola
    -0.75
    sample
    -0.71
    quartered
    -0.70
    adish
    -0.69
     Origin
    -0.69
    Redditor
    -0.69
    Experience
    -0.68
    illusion
    -0.67
    Production
    -0.67
    payment
    -0.66
    POSITIVE LOGITS
     unsuccessfully
    0.99
     harder
    0.77
     desperately
    0.77
     Faul
    0.71
     tried
    0.69
    endants
    0.65
     contradict
    0.64
    unta
    0.64
     Pett
    0.64
     valiant
    0.63
    Activations Density 0.045%

    No Known Activations