INDEX

Explanations

the phrase "Read more" typically found in an online article to indicate further information is available

oai_token-act-pair · gpt-3.5-turbo

references to reading or accessing further information

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

 Reborn

-0.64

STAR

-0.59

.--

-0.59

poon

-0.57

intend

-0.57

naire

-0.55

 âĢķ

-0.54

thro

-0.54

 misunder

-0.54

 ......

-0.54

POSITIVE LOGITS

iesta

0.72

arching

0.67

iness

0.64

 More

0.62

alys

0.62

 more

0.61

edy

0.58

ileged

0.58

More

0.57

etition

0.56

Activations Density 0.014%

the phrase "Read more" typically found in an online article to indicate further information is available

references to reading or accessing further information

No Comments

No Known Activations