INDEX

Explanations

negative words and phrases

oai_token-act-pair · gpt-3.5-turbo

sentiments of irony and surprise in the text

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

»Ĵ

-0.74

ãĤ¼ãĤ¦ãĤ¹

-0.67

irie

-0.67

 warr

-0.62

ebus

-0.60

eteria

-0.60

gang

-0.60

iami

-0.60

ãĥīãĥ©

-0.58

usha

-0.58

POSITIVE LOGITS

 considering

1.42

especially

1.32

 especially

1.29

 because

1.22

 given

1.19

 since

1.16

particularly

1.15

 particularly

1.14

because

1.13

given

1.13

Activations Density 0.271%

negative words and phrases

sentiments of irony and surprise in the text

No Comments

No Known Activations