INDEX

Explanations

mathematical expressions and symbols, particularly involving display styles

oai_token-act-pair · gpt-3.5-turbo

sequences related to mathematical notation or formatting

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

Ago

-0.69

Downloadha

-0.69

 sweats

-0.66

 Aware

-0.65

 Shroud

-0.63

 Niet

-0.63

ascus

-0.62

 Antiqu

-0.62

Reloaded

-0.62

 Notting

-0.61

POSITIVE LOGITS

frac

1.35

display

1.14

text

1.10

ŃĶ

1.06

Delta

1.00

begin

0.98

align

0.96

circ

0.94

cal

0.92

rm

0.92

Activations Density 0.019%

mathematical expressions and symbols, particularly involving display styles

sequences related to mathematical notation or formatting

No Comments

No Known Activations