INDEX

Explanations

phrases indicating a call for action

oai_token-act-pair · gpt-3.5-turbo

repetitive phrases emphasizing personal perspective or requests for feedback

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

ibles

-0.76

ebus

-0.76

pires

-0.73

Downloadha

-0.66

pired

-0.66

ORED

-0.65

lived

-0.64

soDeliveryDate

-0.63

NAS

-0.62

tained

-0.62

POSITIVE LOGITS

 assure

1.16

 explain

1.15

 clarify

1.11

 introduce

1.10

 tell

1.08

 know

1.08

 remind

1.06

 reiterate

1.05

 reassure

1.04

 guess

1.03

Activations Density 0.028%

phrases indicating a call for action

repetitive phrases emphasizing personal perspective or requests for feedback

No Comments

No Known Activations