INDEX

Explanations

phrases related to meeting a certain standard or expectation

oai_token-act-pair · gpt-3.5-turbo

references to a mark or notable achievement

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

ILLE

-0.78

æ©Ł

-0.67

amily

-0.63

foreseen

-0.60

 traged

-0.59

eger

-0.59

 Franch

-0.59

amera

-0.59

aband

-0.58

 pand

-0.58

POSITIVE LOGITS

manship

1.32

emark

1.00

downs

0.97

down

0.97

 Twain

0.83

ename

0.83

marks

0.81

ups

0.79

posts

0.75

vich

0.73

Activations Density 0.020%

phrases related to meeting a certain standard or expectation

references to a mark or notable achievement

No Comments

No Known Activations