INDEX

Explanations

instances of time-related phrases, particularly mentioning the number of days

oai_token-act-pair · gpt-3.5-turbo

mentions of time durations, specifically the word "days."

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

wig

-0.72

allo

-0.66

uana

-0.63

fan

-0.63

olph

-0.62

bender

-0.62

opter

-0.61

rey

-0.61

peer

-0.60

 Proceed

-0.60

POSITIVE LOGITS

 days

3.66

 Days

2.51

 weeks

2.46

Days

2.22

 months

2.22

 nights

2.16

 hours

2.12

days

1.96

 years

1.78

 mornings

1.78

Activations Density 0.027%

instances of time-related phrases, particularly mentioning the number of days

mentions of time durations, specifically the word "days."

No Comments

No Known Activations