INDEX

Explanations

phrases indicating superlatives or rankings

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

pmwiki

-0.76

hyde

-0.73

bleacher

-0.65

uyomi

-0.59

 Cosponsors

-0.58

yna

-0.57

 Unsure

-0.57

BAT

-0.57

ouched

-0.56

gow

-0.56

POSITIVE LOGITS

 theirs

0.95

0.93

0.90

0.88

all

0.84

any

0.83

 hers

0.82

0.80

Activations Density 0.062%

phrases indicating superlatives or rankings

No Comments

No Known Activations