INDEX

Explanations

phrases related to working and employment

oai_token-act-pair · gpt-3.5-turbo

punctuation marks

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

ython

-0.65

isi

-0.62

eln

-0.59

iya

-0.59

aurus

-0.59

obia

-0.57

arah

-0.57

 Secrets

-0.56

"!

-0.56

hib

-0.56

POSITIVE LOGITS

 albeit

0.83

um

0.73

huh

0.70

 taxp

0.69

 whereas

0.68

 respectively

0.68

Balt

0.67

uh

0.67

 supra

0.65

but

0.61

Activations Density 0.264%

phrases related to working and employment

punctuation marks

No Comments

No Known Activations