© 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. 7-RES-JB
    11. 17852
    Prev
    Next
    INDEX
    Explanations

    verbs related to pursuing or hunting

    oai_token-act-pair · gpt-3.5-turbo

    instances of the word "chase" and its variations

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 7-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    7
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    animous
    -0.87
    uary
    -0.79
    osate
    -0.77
    brance
    -0.72
    galitarian
    -0.71
    ynski
    -0.68
    reddits
    -0.66
    ifications
    -0.66
    icka
    -0.65
    activation
    -0.65
    POSITIVE LOGITS
     chase
    0.88
    crow
    0.87
     pursu
    0.76
     Phant
    0.76
    eers
    0.76
     chasing
    0.75
    ttes
    0.73
    backs
    0.72
    rily
    0.69
    =-=-=-=-=-=-=-=-
    0.66
    Activations Density 0.050%

    No Known Activations