INDEX

Explanations

adjectives or phrases related to abstract qualities or characteristics

oai_token-act-pair · gpt-3.5-turbo

phrases that convey a sense of curiosity or wonder about various experiences

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

enting

-0.72

eri

-0.72

chool

-0.72

DOS

-0.71

visors

-0.71

ento

-0.70

ESE

-0.69

assisted

-0.69

 Examples

-0.68

asser

-0.66

POSITIVE LOGITS

 uncanny

1.07

 else

1.07

 profoundly

0.93

 eerie

0.92

 deeper

0.91

 liberating

0.91

 intangible

0.91

 deeply

0.89

 peculiar

0.89

 unsettling

0.88

Activations Density 0.058%

adjectives or phrases related to abstract qualities or characteristics

phrases that convey a sense of curiosity or wonder about various experiences

No Comments

No Known Activations