INDEX

Explanations

time durations expressed in months or years

oai_token-act-pair · gpt-3.5-turbo

references to durations or time periods

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

odder

-0.75

âĨĳ

-0.72

nces

-0.68

behind

-0.67

enf

-0.66

itia

-0.64

cci

-0.64

Inside

-0.63

Tomorrow

-0.60

Behind

-0.60

POSITIVE LOGITS

 period

1.32

 span

1.19

period

0.95

 millennia

0.94

decade

0.93

 timeframe

0.89

 prolonged

0.88

 weekend

0.87

 Period

0.87

 lifespan

0.86

Activations Density 0.067%

time durations expressed in months or years

references to durations or time periods

No Comments

No Known Activations