INDEX

Explanations

phrases related to highlighting or emphasizing something

oai_token-act-pair · gpt-3.5-turbo

instances of the word "shine" and its variations, indicating brightness or illumination

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

essor

-0.74

assian

-0.67

ilage

-0.66

agara

-0.63

ãĥ¯ãĥ³

-0.63

population

-0.62

mx

-0.61

agus

-0.61

apist

-0.60

alian

-0.60

POSITIVE LOGITS

 shone

1.14

 brightly

1.13

 brighter

1.09

 brightest

1.08

 shines

1.06

 shine

0.98

lights

0.93

 shining

0.87

shine

0.85

 illuminate

0.85

Activations Density 0.077%

phrases related to highlighting or emphasizing something

instances of the word "shine" and its variations, indicating brightness or illumination

No Comments

No Known Activations