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. 4-RES-JB
    6. 3585
    Prev
    Next
    INDEX
    Explanations

    mentions of secretive or possibly illegal activities, such as financial deals or hidden information about individuals' whereabouts

    oai_token-act-pair · gpt-3.5-turbo

    references to dealings and interactions, particularly in a political or investigative context

    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
    proof
    -0.76
     horm
    -0.67
     reinforcement
    -0.63
    ©¶æ
    -0.62
    prison
    -0.61
     Fulton
    -0.60
    Effective
    -0.59
    condition
    -0.59
    enough
    -0.59
    IENCE
    -0.58
    POSITIVE LOGITS
    cape
    1.12
     dealings
    0.87
    ongs
    0.86
    hip
    0.84
    poons
    0.83
    uits
    0.83
    heet
    0.82
    creen
    0.82
    pring
    0.81
    leeve
    0.81
    Activations Density 0.100%

    No Known Activations