INDEX

Explanations

words related to hidden or secretive actions

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_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

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

 Fathers

-0.67

enegger

-0.67

 Errors

-0.65

 Sonia

-0.65

 Luther

-0.64

 Merchants

-0.64

ortium

-0.64

Pwr

-0.63

 Breaker

-0.63

ãģ®éŃĶ

-0.62

POSITIVE LOGITS

iously

0.98

ormal

0.91

ker

0.88

ping

0.86

kers

0.85

cher

0.82

chers

0.82

some

0.81

ously

0.80

ched

0.80

Activations Density 0.036%

words related to hidden or secretive actions

No Comments

No Known Activations