INDEX

Explanations

phrases indicating taking action or consideration

oai_token-act-pair · gpt-3.5-turbo

commands or directives to consider or take action on specific topics

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

éĹ

-0.72

-0.71

emale

-0.68

cffffcc

-0.68

ãĤ´

-0.67

interstitial

-0.64

åĤ

-0.64

*/(

-0.64

DIR

-0.64

æĹ

-0.63

POSITIVE LOGITS

 Take

0.84

 Consider

0.80

 Your

0.78

resa

0.78

Plus

0.77

 Examples

0.75

Take

0.74

give

0.71

 Think

0.70

 Differences

0.70

Activations Density 0.073%

phrases indicating taking action or consideration

commands or directives to consider or take action on specific topics

No Comments

No Known Activations