INDEX

Explanations

words starting with "By" followed by a person's name or title

oai_token-act-pair · gpt-3.5-turbo

instances of authorship or attribution in the text

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

itives

-0.73

 pains

-0.71

SpaceEngineers

-0.67

paren

-0.64

 joints

-0.64

isable

-0.63

ties

-0.63

hement

-0.62

 manif

-0.62

 cones

-0.61

POSITIVE LOGITS

akuya

0.89

 virtue

0.81

catch

0.79

stand

0.74

 contrast

0.73

pass

0.72

cloneembedreportprint

0.68

ron

0.65

 Rating

0.64

humans

0.63

Activations Density 0.022%

words starting with "By" followed by a person's name or title

instances of authorship or attribution in the text

No Comments

No Known Activations