INDEX

Explanations

words related to chronology and events that happened in the past

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_fs12288-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

12,288

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Sylvia

-0.72

 Plato

-0.70

 solicitation

-0.68

 Solitaire

-0.67

 Cambodia

-0.65

 recip

-0.63

pta

-0.62

 Scrib

-0.62

ICLE

-0.62

 Craigslist

-0.62

POSITIVE LOGITS

bodied

1.16

scale

1.14

eyed

1.07

enough

1.06

sized

1.02

dollar

0.95

ticket

0.95

capacity

0.93

city

0.92

armed

0.91

Activations Density 0.108%

words related to chronology and events that happened in the past

No Comments

No Known Activations