INDEX

Explanations

phrases related to making conclusions or emphasizing points

oai_token-act-pair · gpt-3.5-turbo

expressions of consequence or emphasis

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

Buyable

-0.75

 Opera

-0.69

BuyableInstoreAndOnline

-0.63

 Palest

-0.63

 Symphony

-0.63

 Tanz

-0.62

 Manhattan

-0.62

 Mansion

-0.61

Gan

-0.61

 Constantin

-0.61

POSITIVE LOGITS

º

1.00

£

0.97

¬

0.97

Ĵ

0.97

¼

0.97

ı

0.92

ĭ

0.91

į

0.90

¹

0.90

¡

0.89

Activations Density 0.067%

phrases related to making conclusions or emphasizing points

expressions of consequence or emphasis

No Comments

No Known Activations