INDEX

Explanations

the name "Alex" with varying levels of importance or context

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

 otherwise

-0.73

recy

-0.66

 prevail

-0.59

 combust

-0.58

 germ

-0.58

 finer

-0.57

itiveness

-0.57

 initiation

-0.57

 altogether

-0.56

liness

-0.56

POSITIVE LOGITS

ei

1.23

andra

1.15

andre

1.11

opoulos

1.05

ey

1.03

iev

0.95

andr

0.95

sand

0.94

ildo

0.92

iov

0.86

Activations Density 0.026%

the name "Alex" with varying levels of importance or context

No Comments

No Known Activations