© 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 Local
    9. 6-RES_SCL-AJT
    11. 32993
    Prev
    Next
    INDEX
    Explanations

    information related to technology, software, and development processes

    oai_token-act-pair · gpt-3.5-turbo
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    neuronpedia/gpt2-small__res_scl-ajt/6-res_scl-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
    674
    +0.13
    0.6%
    7
    +0.12
    0.5%
    372
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    14
    +0.13
    0.08
    297
    +0.12
    0.10
    441
    +0.10
    0.07
    Negative Logits
    buster
    -0.79
     sucks
    -0.63
    breaker
    -0.63
    antis
    -0.63
    strip
    -0.63
    bowl
    -0.62
     {"
    -0.61
     Chains
    -0.61
    icles
    -0.60
    });
    -0.60
    POSITIVE LOGITS
     plenty
    0.94
     unanim
    0.85
     overlap
    0.82
     indications
    0.81
     disagreement
    0.81
     ambiguity
    0.80
     possibility
    0.79
     differences
    0.77
     LOT
    0.75
     precedent
    0.75
    Activations Density 9.916%

    No Known Activations