INDEX

Explanations

verbs indicating continuation or extension of a previous thought or action

oai_token-act-pair · gpt-3.5-turbo

verbs indicating ongoing or repetitive actions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 transformative

-0.62

helps

-0.56

 harms

-0.53

 Workers

-0.53

aples

-0.51

 beware

-0.51

 Gifts

-0.50

chrome

-0.50

stand

-0.50

 hires

-0.49

POSITIVE LOGITS

 rhet

0.76

Laughs

0.73

_.

0.68

ONSORED

0.66

laugh

0.65

Rh

0.64

 quoting

0.64

 diplom

0.64

laughs

0.63

 sarcast

0.63

Activations Density 0.120%

verbs indicating continuation or extension of a previous thought or action

verbs indicating ongoing or repetitive actions

No Comments

No Known Activations