© 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. 8-RES-JB
    11. 22496
    Prev
    Next
    INDEX
    Explanations

    phrases related to movement or relocation

    oai_token-act-pair · gpt-3.5-turbo

    phrases indicating exclusion or being removed from a group or place

    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
     Cosponsors
    -0.77
    £ı
    -0.77
    reddits
    -0.72
    NET
    -0.62
     Helpful
    -0.61
    icion
    -0.61
    accompan
    -0.61
     Pwr
    -0.61
    ancial
    -0.60
    uid
    -0.60
    POSITIVE LOGITS
     bounds
    1.16
     existence
    0.85
     hiber
    0.80
     nowhere
    0.80
     sight
    0.78
     harms
    0.77
     shape
    0.76
     sync
    0.73
     course
    0.72
    rosse
    0.70
    Activations Density 0.062%

    No Known Activations