© 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. 10-RES-JB
    11. 24469
    Prev
    Next
    INDEX
    Explanations

    mentions of individuals named David

    oai_token-act-pair · gpt-3.5-turbo

    references to individuals named David

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 10-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    10
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    NetMessage
    -0.98
    cffffcc
    -0.96
    Reply
    -0.83
    nces
    -0.75
    ttes
    -0.75
    phies
    -0.75
    pmwiki
    -0.72
    bably
    -0.71
    ments
    -0.70
    ĻĤ
    -0.70
    POSITIVE LOGITS
     Bowie
    1.10
     Rockefeller
    1.00
     Ic
    0.98
     Beckham
    0.94
     Petraeus
    0.94
     Axel
    0.93
     Cameron
    0.91
    son
    0.91
     Horowitz
    0.88
     Foster
    0.87
    Activations Density 0.024%

    No Known Activations