© 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. 4-RES-JB
    11. 10074
    Prev
    Next
    INDEX
    Explanations

    uses of the phrase "needs to" followed by an action

    oai_token-act-pair · gpt-3.5-turbo

    phrases or constructions indicating necessary actions or requirements

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

    No Comments

    Negative Logits
     Guilty
    -0.76
     Onion
    -0.74
    quickShipAvailable
    -0.72
    aroo
    -0.67
     Compass
    -0.67
    hot
    -0.66
    lip
    -0.64
     Porn
    -0.64
    hots
    -0.63
    ById
    -0.63
    POSITIVE LOGITS
     keep
    1.10
     fulfill
    1.07
     satisfy
    1.06
     clarify
    1.04
     solve
    1.02
     settle
    1.02
     establish
    1.01
     prove
    1.01
     make
    0.99
     overcome
    0.99
    Activations Density 0.094%

    No Known Activations