Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    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. 24537
    Prev
    Next
    INDEX
    Explanations

    phrases indicating a team or individual has not been defeated in a given context

    oai_token-act-pair · gpt-3.5-turbo

    terms related to being unbeaten or undefeated in competitions

    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
     grav
    -0.89
     Sailor
    -0.72
    anger
    -0.68
    Dial
    -0.67
     attachments
    -0.65
    loe
    -0.65
    ner
    -0.65
     Console
    -0.62
    material
    -0.61
     environments
    -0.61
    POSITIVE LOGITS
     unbeaten
    3.77
     undefeated
    3.50
     unbeat
    1.63
     victorious
    1.46
     invincible
    1.37
    efeated
    1.37
     dynasty
    1.23
     coached
    1.15
     reigning
    1.14
     unstoppable
    1.06
    Activations Density 0.036%

    No Known Activations