© 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. 24009
    Prev
    Next
    INDEX
    Explanations

    names of law firms and financial partners

    oai_token-act-pair · gpt-3.5-turbo

    terms related to law firms and partnerships

    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
    soDeliveryDate
    -0.99
     subreddit
    -0.88
    obyl
    -0.85
    Synopsis
    -0.84
    jriwal
    -0.77
     "$:/
    -0.77
    tube
    -0.74
    redd
    -0.71
    externalActionCode
    -0.70
     blackout
    -0.70
    POSITIVE LOGITS
     LLP
    1.36
     Partners
    1.15
     Greenberg
    1.13
     Associates
    1.13
     Perkins
    1.11
     consultancy
    1.08
     Consulting
    1.07
     Advis
    1.07
     Management
    1.06
     consulting
    1.05
    Activations Density 0.230%

    No Known Activations