INDEX

Explanations

links prompting to read more content or additional information

oai_token-act-pair · gpt-3.5-turbo

instances of the phrase "you can read" along with related content

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

BuyableInstoreAndOnline

-0.73

Lenin

-0.68

akuya

-0.63

aired

-0.61

 enthusi

-0.59

ĸļ

-0.59

TPPStreamerBot

-0.59

pires

-0.58

 Attention

-0.58

rang

-0.58

POSITIVE LOGITS

 yourself

0.82

 yourselves

0.79

 yours

0.75

SOURCE

0.74

 plenty

0.71

 easily

0.69

 your

0.69

 Attribution

0.65

 HERE

0.65

owan

0.64

Activations Density 0.080%

links prompting to read more content or additional information

instances of the phrase "you can read" along with related content

No Comments

No Known Activations