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. 24405
    Prev
    Next
    INDEX
    Explanations

    phrases related to accidents involving derailments or actions meant to hinder or harm something

    oai_token-act-pair · gpt-3.5-turbo

    instances of the word "derail" and its variations, as well as any occurrences of the word "sabotage."

    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
    arden
    -0.70
     administered
    -0.66
    rix
    -0.65
    atos
    -0.63
    addy
    -0.63
     recogn
    -0.62
     Hispanic
    -0.62
    Ps
    -0.61
     Palm
    -0.60
    elman
    -0.60
    POSITIVE LOGITS
     derail
    4.08
     derailed
    2.53
     sabotage
    1.34
     sabot
    1.21
     torpedo
    1.12
     rupture
    1.08
     Amtrak
    1.05
     topple
    1.01
     wreck
    0.98
     brake
    0.97
    Activations Density 0.014%

    No Known Activations