© 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. 5289
    Prev
    Next
    INDEX
    Explanations

    references to fantasy elements or characters

    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
    738
    +0.14
    0.6%
    373
    +0.13
    0.6%
    419
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    701
    +0.14
    0.01
    373
    +0.13
    -0.01
    232
    +0.11
    0.02
    Negative Logits
     elbows
    -0.83
    ibaba
    -0.76
     boycot
    -0.73
     boycott
    -0.71
     whine
    -0.70
    bernatorial
    -0.70
    jad
    -0.69
     Pwr
    -0.68
     bandwagon
    -0.68
    hower
    -0.68
    POSITIVE LOGITS
     Mysteries
    1.17
     Grimoire
    1.01
     cryptic
    0.97
    spell
    0.97
    potion
    0.96
     mysteries
    0.95
     secrets
    0.95
     Secrets
    0.93
     Sorceress
    0.90
     runes
    0.88
    Activations Density 0.068%

    No Known Activations