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

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Joseph Bloom · Open Source Sparse Autoencoders for all Residual Stream Layers of GPT2-Small
    5. GPT2-Small
    7. Residual Stream
    9. 11-RES-JB
    11. 7734
    Prev
    Next
    INDEX
    Explanations

    verbs related to explanations or statements made by someone

    oai_token-act-pair · gpt-3.5-turbo

    instances of the verb "went" and its variations, indicating a focus on actions or movements

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 11-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.hook_resid_pre
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    24,576
    Data Type
    torch.float32
    Hook Point
    blocks.11.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    11
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    likely
    -0.63
    eers
    -0.60
    wear
    -0.58
     expires
    -0.58
    auga
    -0.57
    itudes
    -0.56
    ivo
    -0.54
     drains
    -0.53
     detriment
    -0.53
     Depot
    -0.53
    POSITIVE LOGITS
    aded
    0.93
     overboard
    0.89
     undercover
    0.88
     ballistic
    0.88
     forth
    0.87
     viral
    0.84
     further
    0.83
     farther
    0.81
    ading
    0.80
     deeper
    0.80
    Activations Density 0.066%

    No Known Activations