INDEX

Explanations

proper nouns related to people, places, and organizations

oai_token-act-pair · gpt-3.5-turbo

references to organizations and groups within a specific context

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

etter

-0.63

nings

-0.63

enegger

-0.61

asio

-0.60

etts

-0.59

istically

-0.59

rooms

-0.58

illon

-0.58

omore

-0.57

liest

-0.57

POSITIVE LOGITS

 metic

0.60

ãģ£

0.53

é»Ĵ

0.51

ãģª

0.50

 Stun

0.50

Ø¯

0.50

ãģ

0.50

åĲ

0.50

ÙĬ

0.49

×©

0.49

Activations Density 0.709%

proper nouns related to people, places, and organizations

references to organizations and groups within a specific context

No Comments

No Known Activations