© 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. 9880
    Prev
    Next
    INDEX
    Explanations

    instances where a person is critiquing or analyzing something, especially pointing out flaws and issues

    oai_token-act-pair · gpt-3.5-turbo

    critical analysis and evaluation of statements or pieces

    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
    earchers
    -0.73
    sbm
    -0.73
    yss
    -0.65
    Newsletter
    -0.62
    ropolitan
    -0.60
    urers
    -0.59
    abases
    -0.58
    negie
    -0.57
     depends
    -0.57
     incent
    -0.56
    POSITIVE LOGITS
     "#
    1.03
    Redditor
    1.02
     himself
    0.83
     hilar
    0.81
     downright
    0.78
     Reddit
    0.76
     whopping
    0.76
     unsuspecting
    0.75
     "âĢ¦
    0.74
     retweet
    0.73
    Activations Density 1.230%

    No Known Activations