INDEX

Explanations

phrases indicating a negative outcome or prohibition

oai_token-act-pair · gpt-3.5-turbo

negative assertions or statements

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Nice

-0.67

quickShipAvailable

-0.66

GGGGGGGG

-0.64

Got

-0.63

Awesome

-0.62

 snipp

-0.62

 millenn

-0.62

GGGG

-0.62

ument

-0.61

calling

-0.61

POSITIVE LOGITS

 hesitate

1.28

 tolerate

1.22

be

1.13

 necessarily

1.10

 interfere

1.06

 allow

1.01

icably

1.00

 suffice

0.98

 impede

0.94

 succumb

0.93

Activations Density 0.089%

phrases indicating a negative outcome or prohibition

negative assertions or statements

No Comments

No Known Activations