INDEX

Explanations

expressions related to continuous or repeated actions or events

oai_token-act-pair · gpt-3.5-turbo

phrases indicating back-to-back sequences or repetitions

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

"]=>

-0.72

edIn

-0.70

natureconservancy

-0.68

é¾įå¥ĳå£«

-0.67

lied

-0.66

 msec

-0.65

vich

-0.64

uyomi

-0.63

 Photographer

-0.61

ocument

-0.61

POSITIVE LOGITS

 basics

0.88

roth

0.71

 forth

0.70

back

0.67

 haunt

0.66

 heels

0.66

 knees

0.65

forth

0.65

arters

0.65

backs

0.65

Activations Density 0.054%

expressions related to continuous or repeated actions or events

phrases indicating back-to-back sequences or repetitions

No Comments

No Known Activations