INDEX

Explanations

phrases related to business and organizational positions

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

-0.66

 pilgrims

-0.60

 rive

-0.60

NetMessage

-0.56

aepernick

-0.56

 greedy

-0.55

lihood

-0.55

pel

-0.55

rums

-0.55

POSITIVE LOGITS

 overseeing

1.04

chief

1.03

 emer

0.99

general

0.96

director

0.90

utive

0.89

ially

0.84

 director

0.84

 GENERAL

0.83

 chair

0.82

Activations Density 3.355%

phrases related to business and organizational positions

No Comments

No Known Activations