INDEX

Explanations

phrases indicating personal perceptions or beliefs

oai_token-act-pair · gpt-3.5-turbo

present tense verbs and forms of "to be" indicating personal identity and existence

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

eger

-0.68

pedia

-0.66

jri

-0.66

 refres

-0.64

aign

-0.63

 Holiday

-0.62

ibaba

-0.60

ysis

-0.60

downs

-0.59

utions

-0.59

POSITIVE LOGITS

nt

1.17

 indeed

0.99

 actually

0.91

 somehow

0.86

 never

0.83

 truly

0.79

 destined

0.77

 incapable

0.76

not

0.75

 indispensable

0.75

Activations Density 0.579%

phrases indicating personal perceptions or beliefs

present tense verbs and forms of "to be" indicating personal identity and existence

No Comments

No Known Activations