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

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI 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. 3-RES-JB
    11. 16737
    Prev
    Next
    INDEX
    Explanations

    terms related to dissolution or disbanding

    oai_token-act-pair · gpt-3.5-turbo

    terms related to dissolution or disbanding in various contexts

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 3-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    3
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    efully
    -0.91
    Reply
    -0.74
    Lear
    -0.71
    Demand
    -0.70
    MAT
    -0.69
    airs
    -0.68
    itudes
    -0.67
    eful
    -0.66
    Orig
    -0.66
    eway
    -0.66
    POSITIVE LOGITS
     dissolved
    1.18
     dissolve
    0.98
     dissolution
    0.93
     disband
    0.88
     defunct
    0.83
    owship
    0.79
     Lumpur
    0.71
     aback
    0.71
     Seym
    0.70
     bankrupt
    0.70
    Activations Density 0.026%

    No Known Activations