INDEX

Explanations

proper nouns and technical terms related to a specific topic or industry

oai_token-act-pair · gpt-3.5-turbo

occurrences of specific capitalized terms or proper nouns

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

 Atlantic

-0.75

 wart

-0.72

 drift

-0.70

chnology

-0.67

sea

-0.66

ength

-0.66

 Glac

-0.65

 Lauder

-0.65

 grounding

-0.65

bilt

-0.61

POSITIVE LOGITS

bing

1.21

bies

1.11

bish

1.10

bers

1.08

ļéĨĴ

1.03

bed

1.00

bles

1.00

bie

0.99

lishing

0.99

lique

0.95

Activations Density 0.015%

proper nouns and technical terms related to a specific topic or industry

occurrences of specific capitalized terms or proper nouns

No Comments

No Known Activations