INDEX

Explanations

phrases related to critical assessments or criticisms

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

 Dish

-0.57

 Armored

-0.56

 Tamil

-0.55

 Heights

-0.55

 Mecca

-0.55

 Erit

-0.55

Khe

-0.55

 favour

-0.54

ONSORED

-0.53

 ensemble

-0.53

POSITIVE LOGITS

abouts

1.64

upon

1.26

after

0.94

fore

0.93

FORE

0.82

etheless

0.78

ngth

0.77

are

0.76

 aren

0.75

with

0.75

Activations Density 0.327%

phrases related to critical assessments or criticisms

No Comments

No Known Activations