© 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. 11-RES-JB
    11. 7107
    Prev
    Next
    INDEX
    Explanations

    phrases related to ensuring compliance and safety

    oai_token-act-pair · gpt-3.5-turbo
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 11-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    11
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    NetMessage
    -0.89
    pmwiki
    -0.84
    ibble
    -0.77
     speculate
    -0.71
    onnaissance
    -0.71
    ulia
    -0.71
     Difficulty
    -0.65
     Finder
    -0.65
     Wonders
    -0.64
    rys
    -0.64
    POSITIVE LOGITS
     properly
    1.28
     adequately
    1.19
     safe
    1.11
     appropriately
    1.05
     compliant
    1.03
     not
    1.02
     complying
    1.01
     respectful
    0.98
     correctly
    0.96
     sufficiently
    0.93
    Activations Density 0.173%

    No Known Activations