© 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. Apollo Research · Taylor · Identifying Functionally Important Features with End-to-End Sparse Dictionary Learning
    5. GPT2-Small
    7. Residual Similar CE E2E Recon
    9. 6-RES_SCEFR-AJT
    11. 3803
    Prev
    Next
    INDEX
    Explanations

    fantasy weapon names and descriptions

    oai_token-act-pair · gpt-3.5-turbo
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    neuronpedia/gpt2-small__res_scefr-ajt/6-res_scefr-ajt
    Prompts (Dashboard)
    12,288 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    46,080
    Data Type
    torch.float32
    Hook Point
    blocks.6.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    apollo-research/Skylion007-openwebtext-tokenizer-gpt2
    Hook Point Layer
    6
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    135
    +0.09
    0.5%
    545
    +0.09
    0.5%
    436
    +0.07
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    708
    +0.09
    0.01
    593
    +0.09
    0.01
    476
    +0.07
    0.01
    Negative Logits
    paces
    -0.76
    ometimes
    -0.68
    structed
    -0.62
    _>
    -0.59
    abilia
    -0.56
    pace
    -0.56
    forum
    -0.56
    urus
    -0.55
    osterone
    -0.55
    anmar
    -0.53
    POSITIVE LOGITS
     enchant
    0.61
     Alchemist
    0.57
     Elemental
    0.54
     rune
    0.52
     Ame
    0.52
    nir
    0.50
     Dance
    0.50
     skulls
    0.50
     Siber
    0.49
     deeds
    0.49
    Activations Density 0.043%

    No Known Activations