INDEX

Explanations

phrases related to various news events and articles

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_fs768-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

768

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

handedly

-0.70

naires

-0.67

bats

-0.66

gran

-0.64

ĪĴ

-0.62

knife

-0.61

nov

-0.61

ogy

-0.60

rams

-0.60

gone

-0.59

POSITIVE LOGITS

 WATCHED

0.93

 Expand

0.92

 Thumbnails

0.88

 VIDEOS

0.84

 Loading

0.84

 toggle

0.83

 caption

0.78

 IMAGES

0.78

 Advertisement

0.77

advertisement

0.77

Activations Density 3.371%

phrases related to various news events and articles

No Comments

No Known Activations