INDEX

Explanations

information related to news content or updates

oai_token-act-pair · gpt-3.5-turbo

repeated mentions of "news" and related terms indicating updates or information

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

ignt

-0.70

Â¯Â¯

-0.69

ovember

-0.67

asus

-0.66

ukong

-0.65

 Cheong

-0.64

 Hamm

-0.63

ueless

-0.63

 Reson

-0.62

 sshd

-0.61

POSITIVE LOGITS

flash

1.01

room

1.00

worthy

0.92

worthiness

0.86

ource

0.83

letters

0.83

rooms

0.80

reader

0.79

letter

0.77

giving

0.76

Activations Density 0.025%

information related to news content or updates

repeated mentions of "news" and related terms indicating updates or information

No Comments

No Known Activations