INDEX

Explanations

references to the TV network CW and mentions of mergers

oai_token-act-pair · gpt-3.5-turbo

references to network television and merger-related topics

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

nesia

-0.76

iquette

-0.74

 beads

-0.66

Kn

-0.65

ele

-0.65

 chefs

-0.64

FREE

-0.64

iam

-0.64

bec

-0.63

 Coco

-0.63

POSITIVE LOGITS

 merger

1.81

CW

1.78

 anomaly

1.62

 anomalies

1.50

 deviation

1.14

usp

1.13

CPC

1.01

uve

0.99

 deviations

0.99

Dud

0.98

Activations Density 0.035%

references to the TV network CW and mentions of mergers

references to network television and merger-related topics

No Comments

No Known Activations