INDEX

Explanations

informative phrases containing instructions or explanations

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate instructional content

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

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

enance

-0.76

oppable

-0.67

enegger

-0.62

threat

-0.62

volent

-0.61

Politics

-0.60

orical

-0.60

Rum

-0.59

AIN

-0.59

orter

-0.58

POSITIVE LOGITS

to

0.92

 toget

0.79

--------------------------------------------------------

0.78

semble

0.78

to

0.75

 easy

0.74

you

0.73

 much

0.72

To

0.71

TO

0.70

Activations Density 0.069%

informative phrases containing instructions or explanations

phrases that indicate instructional content

No Comments

No Known Activations