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. 17093
    Prev
    Next
    INDEX
    Explanations

    adjectives related to connection or linkage

    oai_token-act-pair · gpt-3.5-turbo

    instances of the word "connecting" and related concepts

    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
     attend
    -0.78
    ãĥīãĥ©
    -0.77
     participate
    -0.76
     behave
    -0.76
     speak
    -0.75
     abandon
    -0.75
     appear
    -0.74
     leave
    -0.73
     buy
    -0.73
     hesitate
    -0.73
    POSITIVE LOGITS
     determining
    2.32
     defining
    2.20
     detecting
    2.05
     assessing
    2.02
     establishing
    1.97
     predicting
    1.96
     connecting
    1.94
     distinguishing
    1.92
     projecting
    1.92
     evaluating
    1.91
    Activations Density 0.052%

    No Known Activations