© 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. OpenAI · Sparse Autoencoder for GPT2-Small - v5 2024
    5. GPT2-Small
    7. Residual Post - 32k
    9. 8-RES_POST_32K-OAI
    11. 4851
    Prev
    Next
    INDEX
    Explanations

    phrases related to arguments and the discourse surrounding them

    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
    331
    +0.12
    0.5%
    556
    +0.10
    0.4%
    401
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    679
    +0.12
    0.13
    377
    +0.10
    0.11
    720
    +0.09
    0.10
    Negative Logits
     destiny
    -0.71
     liaison
    -0.65
    Shape
    -0.65
     chang
    -0.64
     chats
    -0.63
     imprint
    -0.62
     births
    -0.62
     Collector
    -0.62
     Serie
    -0.62
     markings
    -0.61
    POSITIVE LOGITS
    Arg
    1.33
    argument
    1.27
     arguments
    1.18
     arguing
    1.18
     Argument
    1.16
     argument
    1.10
     assertions
    1.08
    Claim
    1.07
     argued
    1.05
     assertion
    1.04
    Activations Density 3.222%

    No Known Activations