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. 10882
    Prev
    Next
    INDEX
    Explanations

    words related to negative emotions or experiences such as 'depressing' or 'disappointing'

    oai_token-act-pair · gpt-3.5-turbo

    themes related to negativity and disappointment

    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
    •layer 0 - sadness

    No Comments

    Negative Logits
    ona
    -0.91
    tree
    -0.80
    ibi
    -0.80
    aldo
    -0.75
    unity
    -0.72
     Immunity
    -0.70
    qa
    -0.69
    onomy
    -0.68
    deen
    -0.67
     Xavier
    -0.67
    POSITIVE LOGITS
     depress
    3.84
     depressing
    1.90
     disappoint
    1.62
     disturb
    1.30
     baff
    1.19
     cush
    1.18
     sadd
    1.18
     crim
    1.14
     subtract
    1.10
     stimul
    0.98
    Activations Density 0.028%

    No Known Activations