© 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. 15723
    Prev
    Next
    INDEX
    Explanations

    unique character sequences that form a distinct pattern

    oai_token-act-pair · gpt-3.5-turbo

    phrases related to legal and procedural contexts

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    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
     cheaply
    -0.72
     hoped
    -0.68
     ze
    -0.68
    odynam
    -0.68
     blame
    -0.67
    omics
    -0.66
     mull
    -0.66
     thanks
    -0.65
    psey
    -0.65
     whit
    -0.65
    POSITIVE LOGITS
    shall
    1.42
     Notwithstanding
    1.38
     ``(
    1.22
    Except
    1.19
    Exception
    1.18
     Definitions
    1.18
    Whoever
    1.17
    Applic
    1.15
     Provided
    1.14
    TIT
    1.13
    Activations Density 0.163%

    No Known Activations