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. 17356
    Prev
    Next
    INDEX
    Explanations

    particular numerical values appearing in a structured format, particularly the number "45"

    oai_token-act-pair · gpt-3.5-turbo

    numerical values associated with statistics or measurements

    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
     shepherd
    -0.72
     funnel
    -0.64
     human
    -0.62
     visitation
    -0.60
     aide
    -0.60
     reign
    -0.60
     trough
    -0.59
     mine
    -0.58
     swirl
    -0.58
    tains
    -0.58
    POSITIVE LOGITS
    45
    3.17
    44
    2.21
    46
    2.12
    55
    2.07
    35
    1.96
    47
    1.96
    43
    1.96
    48
    1.91
    40
    1.89
    65
    1.89
    Activations Density 0.043%

    No Known Activations