INDEX

Explanations

times and dates mentioned in a specific format

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

gettable

-0.72

proof

-0.71

itably

-0.68

flies

-0.68

istically

-0.68

bee

-0.68

izations

-0.67

izable

-0.66

hist

-0.66

izoph

-0.65

POSITIVE LOGITS

EDT

0.87

EMBER

0.87

DAQ

0.86

CET

0.77

ION

0.76

OUR

0.75

Ãī

0.74

ember

0.72

ARGET

0.72

CHR

0.72

Activations Density 0.014%

times and dates mentioned in a specific format

No Comments

No Known Activations