© 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. 4-RES-JB
    11. 14825
    Prev
    Next
    INDEX
    Explanations

    contractions of the verb "to be" followed by high activation values

    oai_token-act-pair · gpt-3.5-turbo

    instances of the contraction "I'd" indicating past intentions or plans

    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
    æ³
    -0.76
     Gutenberg
    -0.76
    PI
    -0.68
    topic
    -0.67
    asse
    -0.66
    çĶŁ
    -0.63
     Diet
    -0.63
    SUP
    -0.63
    ISO
    -0.62
    çİĭ
    -0.62
    POSITIVE LOGITS
     been
    0.83
     gotta
    0.83
    hran
    0.83
     gladly
    0.79
     gotten
    0.77
    escription
    0.77
     gone
    0.74
    hap
    0.70
    emonium
    0.69
    anoia
    0.68
    Activations Density 0.018%

    No Known Activations