INDEX

Explanations

mentions of specific locations or businesses by name

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_fs3072-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

3,072

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

uracy

-0.73

 Lauder

-0.72

 Leone

-0.66

OPE

-0.66

 Seconds

-0.65

AUT

-0.65

âķĲâķĲ

-0.62

each

-0.61

 curled

-0.59

ossession

-0.59

POSITIVE LOGITS

stay

1.56

tenance

1.39

deck

1.02

 arteries

1.01

frame

1.01

stream

1.00

 artery

0.99

boards

0.95

 culprit

0.92

board

0.89

Activations Density 0.298%

mentions of specific locations or businesses by name

No Comments

No Known Activations