INDEX

Explanations

insight or information

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

ategory

-0.67

Fax

-0.64

enegger

-0.63

 Agric

-0.61

 Cancel

-0.60

senal

-0.55

 Again

-0.55

Rez

-0.54

Ern

-0.53

HAM

-0.53

POSITIVE LOGITS

 into

1.30

 Into

1.03

 glean

1.02

into

1.01

 INTO

1.01

ively

1.00

spection

0.93

ibly

0.92

 insights

0.90

orial

0.89

Activations Density 0.047%

insight or information

No Comments

No Known Activations