INDEX

Explanations

phrases related to extremes or polar opposites

oai_token-act-pair · gpt-3.5-turbo

words that indicate extreme or pivotal concepts, particularly related to epitomes and antitheses

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 Bucks

-0.67

 guiActiveUnfocused

-0.67

hower

-0.62

 Franklin

-0.62

BSD

-0.61

 downhill

-0.61

CDC

-0.60

 Palmer

-0.60

WHERE

-0.60

HQ

-0.60

POSITIVE LOGITS

hetically

1.11

hetical

1.01

aph

0.98

viation

0.97

helial

0.97

orical

0.92

oria

0.91

hetic

0.91

het

0.89

ophon

0.88

Activations Density 0.017%

phrases related to extremes or polar opposites

words that indicate extreme or pivotal concepts, particularly related to epitomes and antitheses

No Comments

No Known Activations