© 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. 20280
    Prev
    Next
    INDEX
    Explanations

    text related to crafting or building things

    oai_token-act-pair · gpt-3.5-turbo

    references to specific individuals or notable figures

    oai_token-act-pair · gpt-4o-miniTriggered by @bot

    proper nouns after verbs

    Explanation Uploaded by User
    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
    •contextual features
    fires on a proper noun after a verb

    No Comments

    Negative Logits
     notor
    -0.67
     incorpor
    -0.61
     reluct
    -0.57
     confir
    -0.57
     prest
    -0.57
     conclud
    -0.55
     denomin
    -0.55
     sugg
    -0.54
     predec
    -0.52
     comr
    -0.52
    POSITIVE LOGITS
     onto
    0.70
    RNA
    0.55
     into
    0.52
     hostage
    0.50
    til
    0.49
     badge
    0.48
    crit
    0.46
     squarely
    0.46
    tones
    0.45
    gently
    0.45
    Activations Density 1.473%

    No Known Activations