INDEX

Explanations

transitional phrases indicating contrast or exceptions in arguments

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs49152-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

49,152

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

QB

-0.75

ãĤ©

-0.74

äºĶ

-0.66

 shortest

-0.66

 cooldown

-0.64

çľ

-0.62

Nation

-0.61

é¾

-0.61

 beginner

-0.60

 stall

-0.60

POSITIVE LOGITS

 nevertheless

1.34

 nonetheless

1.22

 anecd

0.86

 luckily

0.80

efficients

0.76

ebus

0.73

poons

0.73

ulia

0.70

netflix

0.70

tsky

0.69

Activations Density 0.273%

transitional phrases indicating contrast or exceptions in arguments

No Comments

No Known Activations