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

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI 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. 11876
    Prev
    Next
    INDEX
    Explanations

    instances of criticism or commentary in text

    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
    393
    +0.30
    1.0%
    546
    +0.14
    0.5%
    192
    +0.14
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    172
    +0.30
    0.09
    192
    +0.14
    0.07
    743
    +0.14
    0.06
    Negative Logits
    eus
    -1.05
    ivas
    -1.03
    tty
    -1.02
    OUR
    -1.00
    IVE
    -0.99
    rise
    -0.98
    Redditor
    -0.97
    roads
    -0.96
    âĹ¼
    -0.93
    erest
    -0.93
    POSITIVE LOGITS
     evidenced
    1.74
     opposed
    1.49
    bestos
    1.47
     well
    1.41
    pired
    1.39
    semb
    1.34
    phy
    1.28
    piring
    1.25
    pires
    1.25
     shown
    1.24
    Activations Density 0.948%

    No Known Activations