INDEX

Explanations

locations or places

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

bal

-0.69

hift

-0.67

quel

-0.65

missions

-0.63

efe

-0.62

icer

-0.62

iasis

-0.61

gotten

-0.61

rett

-0.61

estamp

-0.60

POSITIVE LOGITS

 else

1.52

abouts

1.21

 Else

1.08

 imaginable

1.07

Else

0.95

else

0.95

 except

0.93

upon

0.83

 Everywhere

0.83

 conceivable

0.83

Activations Density 0.020%

locations or places

No Comments

No Known Activations