© 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. 4-RES-JB
    11. 22013
    Prev
    Next
    INDEX
    Explanations

    references related to capacity, particularly in terms of volume or capability

    oai_token-act-pair · gpt-3.5-turbo

    references to numerical or quantitative capacity

    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
    roma
    -0.81
    aida
    -0.74
    rom
    -0.72
    dash
    -0.71
    hof
    -0.67
    bara
    -0.67
    generic
    -0.67
    ograph
    -0.66
    chy
    -0.65
    enegger
    -0.64
    POSITIVE LOGITS
     capacities
    1.28
     capacity
    1.24
     utilization
    1.04
     Capacity
    0.95
    ibilities
    0.93
     mismatch
    0.87
    capacity
    0.82
    encies
    0.78
     gaps
    0.78
    auga
    0.77
    Activations Density 0.017%

    No Known Activations