INDEX

Explanations

years from the 1990s

oai_token-act-pair · gpt-3.5-turbo

mentions of the year 1997

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

hop

-0.69

Tu

-0.67

hatt

-0.63

Tomorrow

-0.63

Nation

-0.62

ship

-0.61

tro

-0.61

ysc

-0.61

rop

-0.61

EStreamFrame

-0.60

POSITIVE LOGITS

ĸļ

0.89

ãĤ¦ãĤ¹

0.86

å¹

0.83

-'

0.83

ãĥĥãĥī

0.82

0.73

 "$:/

0.71

 onward

0.70

0.68

Activations Density 0.025%

years from the 1990s

mentions of the year 1997

No Comments

No Known Activations