INDEX

Explanations

the word "trim" at various activation levels

oai_token-act-pair · gpt-3.5-turbo

the word "trim."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

founded

-0.95

ordial

-0.68

Haunted

-0.66

 Quake

-0.64

Aw

-0.63

geist

-0.62

 alienation

-0.62

izens

-0.61

cloud

-0.60

 Guilty

-0.60

POSITIVE LOGITS

 trim

3.94

 trimmed

2.49

 shave

1.43

 tidy

1.23

 shaved

1.22

 trunc

1.18

 shaving

1.14

 slim

1.10

 clipping

1.08

 haircut

1.07

Activations Density 0.019%

the word "trim" at various activation levels

the word "trim."

No Comments

No Known Activations