INDEX

Explanations

words related to locations or names ending in "-ar"

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "ar."

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

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

¬¼

-0.85

er

-0.70

éĹĺ

-0.69

 depos

-0.69

ĨĴ

-0.68

ĸļ

-0.67

eur

-0.66

-0.65

assetsadobe

-0.62

ĪĴ

-0.62

POSITIVE LOGITS

aja

1.06

ashtra

1.05

riors

1.05

thur

1.03

ctica

0.99

ifle

0.98

acter

0.96

onson

0.95

raf

0.95

ayan

0.95

Activations Density 0.061%

words related to locations or names ending in "-ar"

occurrences of the word "ar."

No Comments

No Known Activations