INDEX

Explanations

issues or complexities within processes or environments

oai_token-act-pair · gpt-3.5-turbo

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

EStream

-0.96

idth

-0.83

ulkan

-0.81

merga

-0.78

wic

-0.78

SHIP

-0.77

ZA

-0.76

ESE

-0.76

heid

-0.76

umbn

-0.74

POSITIVE LOGITS

 unpredictable

1.22

 predictable

1.11

 informative

1.08

 manageable

1.06

 therefore

1.03

 devoid

1.02

 enjoyable

1.01

 unfor

1.01

 insightful

0.99

 inaccessible

0.99

Activations Density 0.167%

issues or complexities within processes or environments

No Comments

No Known Activations