INDEX

Explanations

phrases containing words related to changes or actions, often with a consequential tone

oai_token-act-pair · gpt-3.5-turbo

commas 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 @ 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

¬¼

-0.66

ulz

-0.64

antz

-0.64

grain

-0.61

hoe

-0.60

oons

-0.60

corn

-0.59

asses

-0.58

cott

-0.58

ge

-0.57

POSITIVE LOGITS

 albeit

1.20

 namely

1.00

 although

0.86

 however

0.84

 though

0.83

 according

0.80

 respectively

0.79

 regardless

0.79

 including

0.79

 barring

0.77

Activations Density 0.492%

phrases containing words related to changes or actions, often with a consequential tone

commas in the text

No Comments

No Known Activations