INDEX

Explanations

proper nouns and geopolitical terms

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_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

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

Ò

-0.80

thood

-0.75

ï¿½ï¿½ï¿½ï¿½

-0.74

ceive

-0.72

eno

-0.70

Ïī

-0.69

imi

-0.68

leeve

-0.67

 without

-0.67

ÏĢ

-0.67

POSITIVE LOGITS

oret

1.63

resa

1.36

odore

1.31

 biggest

1.25

ories

1.21

 easiest

1.19

 downside

1.19

 latter

1.18

 simplest

1.18

 earliest

1.11

Activations Density 2.264%

proper nouns and geopolitical terms

No Comments

No Known Activations