INDEX

Explanations

phrases containing the word "are"

oai_token-act-pair · gpt-3.5-turbo

present tense forms of the verb "to be."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

rouse

-0.72

odor

-0.71

 emerges

-0.65

 disappears

-0.63

odder

-0.63

ademic

-0.62

 mater

-0.62

icist

-0.61

 entails

-0.60

 occurs

-0.59

POSITIVE LOGITS

 thankful

1.10

 glad

1.05

 gonna

1.04

 fortunate

1.03

 grateful

0.98

 ourselves

0.97

 proud

0.94

 indebted

0.92

 sorry

0.89

 hoping

0.89

Activations Density 0.154%

phrases containing the word "are"

present tense forms of the verb "to be."

No Comments

No Known Activations