INDEX

Explanations

the prefix 'non-' that indicates negation or absence

oai_token-act-pair · gpt-3.5-turbo

the term "non" in various contexts

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

 Tycoon

-1.01

 Showdown

-0.81

 Cobra

-0.80

 Oaks

-0.75

 McGee

-0.75

 Orchestra

-0.75

 Sparrow

-0.74

 Plate

-0.74

 Banner

-0.73

hower

-0.73

POSITIVE LOGITS

etheless

1.13

chal

1.04

fiction

1.01

epad

0.99

zero

0.99

aligned

0.97

profit

0.95

availability

0.95

linear

0.95

basic

0.94

Activations Density 0.032%

the prefix 'non-' that indicates negation or absence

the term "non" in various contexts

No Comments

No Known Activations