INDEX

Explanations

the word 'Wake', potentially looking for names or locations containing this term

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

ahime

-0.85

otomy

-0.76

mentation

-0.73

afort

-0.71

served

-0.71

matically

-0.71

atri

-0.70

eways

-0.69

uana

-0.68

ioned

-0.67

POSITIVE LOGITS

 Coastal

0.94

wash

0.92

 Hurricanes

0.86

 Herald

0.79

 Beach

0.79

omore

0.78

 Carolina

0.74

 Orange

0.71

gate

0.68

 Wake

0.68

Activations Density 0.034%

the word 'Wake', potentially looking for names or locations containing this term

No Comments

No Known Activations