© 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. 14687
    Prev
    Next
    INDEX
    Explanations

    phrases indicating a task or action to be done

    oai_token-act-pair · gpt-3.5-turbo

    phrases emphasizing the concept of "doing it" or repeating actions

    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
     Tru
    -0.73
     fears
    -0.70
     Flavoring
    -0.69
     Sett
    -0.67
     Strongh
    -0.64
     Objects
    -0.64
     Leader
    -0.64
    Whe
    -0.64
    Parent
    -0.64
    Magikarp
    -0.63
    POSITIVE LOGITS
     job
    0.74
     injustice
    0.74
     justice
    0.73
     thing
    0.72
     experiment
    0.71
    pez
    0.71
    ña
    0.70
    lez
    0.69
     depends
    0.69
    ibaba
    0.67
    Activations Density 0.094%

    No Known Activations