© 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. 8873
    Prev
    Next
    INDEX
    Explanations

    phrases indicating a change or habit from the past

    oai_token-act-pair · gpt-3.5-turbo

    phrases that contain the words "to be" or "to say" in various contexts

    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
    atform
    -0.76
    emetery
    -0.75
    mble
    -0.73
    irement
    -0.69
    daughter
    -0.69
    omm
    -0.68
    ufact
    -0.67
    ģĸ
    -0.66
    ensable
    -0.66
    ellig
    -0.64
    POSITIVE LOGITS
     sparing
    0.85
     liber
    0.76
     interchange
    0.71
    ãĥīãĥ©ãĤ´ãĥ³
    0.67
     levers
    0.65
    CLE
    0.65
     metaphors
    0.64
     tools
    0.64
     cheat
    0.64
     extensively
    0.63
    Activations Density 0.530%

    No Known Activations