© 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. 9403
    Prev
    Next
    INDEX
    Explanations

    specific phrases or terms related to factual information or historical events

    oai_token-act-pair · gpt-3.5-turbo

    references to significant messages, facts, or conditions related to social issues and equality

    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
    ãĥĥãĤ¯
    -0.60
     hemor
    -0.56
     cryst
    -0.55
    ãĥĩãĤ£
    -0.52
     functioning
    -0.51
     disadvant
    -0.51
    BUT
    -0.51
    ãģį
    -0.50
    sing
    -0.49
     msec
    -0.49
    POSITIVE LOGITS
     requires
    1.25
     brings
    1.13
     entails
    1.13
     reveals
    1.10
     involves
    1.08
     gives
    1.06
     reminds
    1.05
     helps
    1.01
     allows
    1.01
     ensures
    1.00
    Activations Density 0.426%

    No Known Activations