© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW
    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    2. OpenAI · Sparse Autoencoder for GPT2-Small - v5 2024
    3. GPT2-Small
    4. Residual Post - 32k
    5. 10-RES_POST_32K-OAI
    6. 29332
    Prev
    Next
    INDEX
    Explanations

    references to specific tasks or entities within the context of various activities or systems

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 10-res_post_32k-oai
    Configuration
    jbloom/GPT2-Small-OAI-v5-32k-resid-post-SAEs/v5_32k_layer_10.pt
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    32,768
    Data Type
    torch.float32
    Hook Name
    blocks.10.hook_resid_post
    Hook Layer
    10
    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
    609
    +0.12
    0.5%
    13
    +0.10
    0.5%
    619
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    314
    +0.12
    -0.01
    288
    +0.10
    0.04
    232
    +0.10
    0.03
    Negative Logits
    oles
    -0.66
    ropolitan
    -0.65
     Va
    -0.62
    éĹĺ
    -0.61
     Lies
    -0.59
    Sales
    -0.58
    ened
    -0.57
     therapists
    -0.57
     psychiatry
    -0.57
     radiator
    -0.57
    POSITIVE LOGITS
    onge
    0.98
    ted
    0.92
    tle
    0.84
    ten
    0.84
    Hub
    0.81
    gin
    0.79
    bones
    0.78
    tw
    0.74
    umbs
    0.74
    take
    0.74
    Activations Density 0.198%

    No Known Activations