© 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. 14851
    Prev
    Next
    INDEX
    Explanations

    comparisons of difficulty or effort

    oai_token-act-pair · gpt-3.5-turbo

    instances of the word "harder" and its associated contexts

    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
     Ut
    -0.63
    opsis
    -0.63
    ifa
    -0.63
     Sources
    -0.62
     Occup
    -0.60
    ********************************
    -0.58
     consultation
    -0.58
     Orig
    -0.57
     eman
    -0.57
    acas
    -0.57
    POSITIVE LOGITS
     harder
    3.45
     tougher
    2.17
     hardest
    2.02
     easier
    1.97
     hard
    1.89
     softer
    1.70
    hard
    1.68
     louder
    1.65
     slower
    1.54
     harsher
    1.53
    Activations Density 0.013%

    No Known Activations