INDEX

Explanations

mention of the word "King" in various contexts

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

TING

-0.73

ted

-0.71

ename

-0.70

ATIONS

-0.69

ters

-0.66

orers

-0.63

ations

-0.62

oresc

-0.62

ATIONAL

-0.61

orescence

-0.59

POSITIVE LOGITS

pin

1.38

pins

1.27

uin

1.23

dom

1.02

doms

1.01

 Kong

1.00

STON

0.92

ston

0.92

 Abdullah

0.91

killer

0.90

Activations Density 0.024%

mention of the word "King" in various contexts

No Comments

No Known Activations