INDEX

Explanations

statements or information about someone receiving credit for accomplishments or contributions

oai_token-act-pair · gpt-3.5-turbo

instances of the word "credited" related to contributions or recognition

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Afgh

-0.71

boats

-0.70

xy

-0.69

 Viet

-0.68

cum

-0.66

eper

-0.63

 tents

-0.62

ago

-0.62

inen

-0.61

angular

-0.60

POSITIVE LOGITS

redited

0.92

 credits

0.89

enza

0.88

 credited

0.88

reon

0.84

heet

0.81

rill

0.79

olor

0.77

iliated

0.76

 Credits

0.74

Activations Density 0.009%

statements or information about someone receiving credit for accomplishments or contributions

instances of the word "credited" related to contributions or recognition

No Comments

No Known Activations