INDEX

Explanations

phrases related to recommendations or suggestions

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs3072-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

3,072

Data Type

torch.float32

Hook Point

blocks.8.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

Fra

-0.74

Wid

-0.63

 Puzzle

-0.63

WI

-0.61

ROS

-0.59

 Syndrome

-0.59

GGGGGGGG

-0.57

 Afgh

-0.57

 Patty

-0.56

 syndrome

-0.56

POSITIVE LOGITS

be

1.19

ered

1.18

nt

1.02

 ideally

1.01

ering

0.95

 strive

0.93

0.92

 aspire

0.91

 beware

0.88

ublic

0.87

Activations Density 0.505%

phrases related to recommendations or suggestions

No Comments

No Known Activations