INDEX

Explanations

emotionally charged adjectives

oai_token-act-pair · gpt-3.5-turbo

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 cycle

-0.67

 GEAR

-0.63

 waivers

-0.62

Origin

-0.61

 minors

-0.61

RI

-0.59

 Korea

-0.58

ok

-0.57

 ALSO

-0.57

 damages

-0.56

POSITIVE LOGITS

acious

1.10

ignant

0.97

entious

0.96

arious

0.93

uous

0.91

romising

0.91

oried

0.89

ious

0.89

worldly

0.87

ithering

0.87

Activations Density 0.268%

emotionally charged adjectives

No Comments

No Known Activations