© 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
    2. Joseph Bloom · Open Source Sparse Autoencoders for all Residual Stream Layers of GPT2-Small
    3. GPT2-Small
    4. Residual Stream
    5. 0-RES-JB
    6. 12172
    Prev
    Next
    INDEX
    Explanations

    words related to physical directions (uphill, downhill) or difficulties/challenges

    oai_token-act-pair · gpt-3.5-turbo

    metaphors related to challenges and struggles

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 0-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    0
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     Fry
    -0.70
    ancies
    -0.70
     Synd
    -0.67
    ry
    -0.67
    omsky
    -0.66
    rit
    -0.65
     Codes
    -0.64
    onom
    -0.64
     Fidel
    -0.62
     Que
    -0.61
    POSITIVE LOGITS
     uphill
    3.61
     downhill
    3.35
     downstream
    1.39
     upstream
    1.17
     paved
    1.12
     snowball
    1.10
    onite
    1.06
     hill
    1.05
     hills
    1.05
     rocky
    1.04
    Activations Density 0.053%

    No Known Activations