© 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
    2. OpenAI · Sparse Autoencoder for GPT2-Small - v5 2024
    3. GPT2-Small
    4. Residual Post - 32k
    5. 8-RES_POST_32K-OAI
    6. 5271
    Prev
    Next
    INDEX
    Explanations

    references to "streets," particularly in the context of protests, gatherings, or social issues

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_post_32k-oai
    Configuration
    jbloom/GPT2-Small-OAI-v5-32k-resid-post-SAEs/v5_32k_layer_8.pt
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    32,768
    Data Type
    torch.float32
    Hook Name
    blocks.8.hook_resid_post
    Hook Layer
    8
    Architecture
    standard
    Context Size
    64
    Dataset
    Skylion007/openwebtext
    Activation Function
    topk
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    420
    +0.13
    0.6%
    365
    +0.11
    0.5%
    748
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    325
    +0.13
    0.03
    288
    +0.11
    0.03
    113
    +0.09
    0.03
    Negative Logits
    iator
    -0.70
    ioned
    -0.68
    estamp
    -0.67
    wcsstore
    -0.67
     takeoff
    -0.66
    izontal
    -0.65
    phabet
    -0.65
    raviolet
    -0.64
    ño
    -0.63
     stove
    -0.63
    POSITIVE LOGITS
    cape
    0.95
     streets
    0.93
    aving
    0.87
     sidewalks
    0.75
    ways
    0.75
    worn
    0.74
    ffiti
    0.74
     halls
    0.73
    fare
    0.73
    ourced
    0.72
    Activations Density 0.148%

    No Known Activations