INDEX

Explanations

phrases related to accountability and responsibility

oai_token-act-pair · gpt-3.5-turbo

statements emphasizing honesty or truthfulness

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

hemer

-0.71

robe

-0.65

ocating

-0.64

igr

-0.63

(?,

-0.63

 Mechdragon

-0.62

 Cruel

-0.60

 Slug

-0.57

 droid

-0.55

ļéĨĴ

-0.55

POSITIVE LOGITS

 rightly

0.95

 doubtless

0.92

 fortunately

0.87

 thankfully

0.83

 reiterate

0.82

 rightfully

0.81

thank

0.81

 reiterated

0.80

 echoed

0.78

withstanding

0.78

Activations Density 0.324%

phrases related to accountability and responsibility

statements emphasizing honesty or truthfulness

No Comments

No Known Activations