INDEX

Explanations

words related to strong opinions or emphasis

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

ioned

-0.77

bourg

-0.71

ulton

-0.71

ulative

-0.71

isson

-0.71

jri

-0.70

engers

-0.70

iem

-0.69

tein

-0.69

NetMessage

-0.69

POSITIVE LOGITS

ove

0.71

reme

0.71

 bananas

0.67

urious

0.67

 nuts

0.67

 fucking

0.66

 delighted

0.65

 Vader

0.65

ĪĴ

0.65

 adore

0.64

Activations Density 0.025%

words related to strong opinions or emphasis

No Comments

No Known Activations