© 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. 3-RES-JB
    11. 11956
    Prev
    Next
    INDEX
    Explanations

    phrases related to specific items or concepts such as "prairie", "microgravity", "elixir", "capitol", "bazaar", "inferno", "phoenix", "zeitgeist", "franca", "monk strap", "vitamins", and "apocalypse"

    oai_token-act-pair · gpt-3.5-turbo

    words related to abstract concepts and transformative experiences

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 3-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    3
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     KR
    -0.74
    NL
    -0.71
    Nap
    -0.70
     Weeks
    -0.68
     Shots
    -0.68
     Others
    -0.68
     RH
    -0.68
    GER
    -0.67
     HS
    -0.67
     SEN
    -0.66
    POSITIVE LOGITS
    tml
    0.71
     appl
    0.71
    ishly
    0.71
    sie
    0.70
    eers
    0.68
    Redditor
    0.68
    olded
    0.68
     hawk
    0.68
    ãĥ³ãĤ¸
    0.67
    ItemTracker
    0.67
    Activations Density 0.179%

    No Known Activations