Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Apollo Research · Taylor · Identifying Functionally Important Features with End-to-End Sparse Dictionary Learning
    3. GPT2-Small
    4. Residual Similar L0 E2E Recon
    5. 2-RES_SLEFR-AJT
    6. 9858
    Prev
    Next
    INDEX
    Explanations

    the word "cliff" (or potential misspellings) in the context of a physical precipice or a metaphorical impending disaster.

    oai_token-act-pair · gemini-1.5-flashTriggered by @johnny

    The token "cliff" appearing in various contexts such as TV show plot summaries, news articles about rescues and accidents, and references to sports, while also being part of locations and as part of political metaphors.

    eleuther_acts_top20 · gpt-3.5-turboTriggered by @johnny

    references to cliffs and related imagery

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    neuronpedia/gpt2-small__res_slefr-ajt/2-res_slefr-ajt
    Prompts (Dashboard)
    12,288 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    46,080
    Data Type
    torch.float32
    Hook Point
    blocks.2.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    apollo-research/Skylion007-openwebtext-tokenizer-gpt2
    Hook Point Layer
    2
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    57
    +0.09
    0.6%
    379
    +0.09
    0.5%
    531
    +0.08
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    288
    +0.09
    0.03
    99
    +0.09
    0.03
    531
    +0.08
    0.02
    Negative Logits
    efully
    -0.71
    eful
    -0.60
    ALD
    -0.56
    ²¾
    -0.54
    wald
    -0.50
     Metatron
    -0.49
     disinfect
    -0.49
     EB
    -0.48
    olith
    -0.47
    ŀ
    -0.47
    POSITIVE LOGITS
    hani
    0.67
     Fighters
    0.54
    cases
    0.53
     Syri
    0.53
    gements
    0.50
    aration
    0.49
     concessions
    0.49
    lycer
    0.49
    antz
    0.49
     sidx
    0.49
    Activations Density 0.071%

    No Known Activations