© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI 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. 7-RES-JB
    11. 13946
    Prev
    Next
    INDEX
    Explanations

    websites

    oai_token-act-pair · gpt-3.5-turbo

    web addresses or URLs

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 7-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    7
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     judgment
    -0.86
     judgments
    -0.84
    ITION
    -0.74
     existence
    -0.73
     proposition
    -0.73
     deportation
    -0.73
     headline
    -0.72
     propositions
    -0.70
     relief
    -0.69
     equival
    -0.69
    POSITIVE LOGITS
    facebook
    1.35
    youtube
    1.25
    example
    1.22
    pn
    1.15
    etsy
    1.15
    imgur
    1.11
    twitch
    1.10
    reddit
    1.07
    daily
    1.06
    esp
    1.06
    Activations Density 0.033%

    No Known Activations