© 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 · Feature Splitting for GPT2-Small
    5. GPT2-Small
    7. Residual Feature Splitting 1536
    9. 8-RES_FS1536-JB
    11. 53
    Prev
    Next
    INDEX
    Explanations

    descriptions of specific abilities and characteristics in a game context

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs1536-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    1,536
    Data Type
    torch.float32
    Hook Point
    blocks.8.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    8
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    326
    +0.17
    0.8%
    282
    +0.10
    0.5%
    747
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    326
    +0.17
    0.14
    738
    +0.10
    0.05
    678
    +0.10
    0.05
    Negative Logits
     Weiner
    -0.84
    mercial
    -0.84
    internet
    -0.82
    çĶ°
    -0.81
    Asia
    -0.77
     Draper
    -0.76
    India
    -0.76
     FedEx
    -0.76
     Exhibit
    -0.74
     ÂŃ
    -0.73
    POSITIVE LOGITS
     teleport
    1.43
     melee
    1.41
     cooldown
    1.40
    Spell
    1.32
     Melee
    1.32
     armor
    1.30
     Spells
    1.27
     Summon
    1.27
     runes
    1.26
     Cooldown
    1.26
    Activations Density 2.864%

    No Known Activations