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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports 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. 8-RES-JB
    11. 6393
    Prev
    Next
    INDEX
    Explanations

    words or phrases related to scolding or reprimanding

    oai_token-act-pair · gpt-3.5-turbo

    instances of the word "sc" and its variations, often in contexts of criticism or confrontation

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    8
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     remembrance
    -0.73
    elsen
    -0.72
     lapse
    -0.69
    heit
    -0.65
    nown
    -0.64
    emonic
    -0.64
    terday
    -0.64
    éĹĺ
    -0.61
     laure
    -0.60
     contender
    -0.60
    POSITIVE LOGITS
    oops
    1.39
    rupulous
    1.38
    ammers
    1.29
    ouring
    1.24
    rawling
    1.24
    atters
    1.22
    opes
    1.20
    ribe
    1.18
    outed
    1.17
    rup
    1.16
    Activations Density 0.011%

    No Known Activations