INDEX

Explanations

requests for information or action using the phrase "Can you" followed by a verb

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.10.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

yk

-0.76

furt

-0.73

DERR

-0.73

ukemia

-0.72

senal

-0.68

stanbul

-0.67

enez

-0.66

hung

-0.66

assic

-0.65

ĸļå£«

-0.65

POSITIVE LOGITS

 imagine

1.29

 afford

1.18

 please

1.16

 PLEASE

1.08

 quantify

1.05

 explain

0.92

 tell

0.91

 conceive

0.90

 assure

0.90

 summarize

0.89

Activations Density 0.043%

requests for information or action using the phrase "Can you" followed by a verb

No Comments

No Known Activations