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. 8-RES-JB
    6. 16415
    Prev
    Next
    INDEX
    Explanations

    phrases related to skills or requirements such as "good to Finals" and "sailor anymore."

    oai_token-act-pair · gpt-3.5-turbo

    phrases that indicate necessity or requirements

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    8
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     unwanted
    -0.69
    JP
    -0.67
    NetMessage
    -0.59
     unwelcome
    -0.57
     Carbuncle
    -0.57
    Title
    -0.55
    Downloadha
    -0.54
    reddits
    -0.53
     Tit
    -0.53
    haps
    -0.53
    POSITIVE LOGITS
     ASAP
    1.09
     Otherwise
    1.01
     lest
    0.99
     beforehand
    0.78
     BEFORE
    0.77
     manually
    0.76
     urgently
    0.73
     Pwr
    0.73
     preferably
    0.73
     somew
    0.70
    Activations Density 0.693%

    No Known Activations