INDEX

Explanations

words indicating responsibility or action

oai_token-act-pair · gpt-3.5-turbo

phrases related to the second person perspective, specifically addressing the reader directly

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

Scroll

-0.65

issues

-0.62

rates

-0.62

ozy

-0.61

idth

-0.60

ournals

-0.58

yrim

-0.58

issue

-0.58

hess

-0.57

 legality

-0.57

POSITIVE LOGITS

 tempted

1.12

 obligated

1.08

 obliged

1.05

 confronted

1.03

 risking

1.03

 able

1.01

 forced

1.00

 rewarded

0.99

 reminded

0.97

 sacrificing

0.96

Activations Density 0.210%

words indicating responsibility or action

phrases related to the second person perspective, specifically addressing the reader directly

No Comments

No Known Activations