INDEX

Explanations

mentions of time specific to the evening

oai_token-act-pair · gpt-3.5-turbo

references to the time of day, specifically "evening"

oai_token-act-pair · gpt-4-turbo

references to "evening."

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

onne

-0.77

eps

-0.77

insured

-0.76

uador

-0.75

Done

-0.75

control

-0.73

ophers

-0.72

hess

-0.72

xon

-0.72

lez

-0.71

POSITIVE LOGITS

 afternoon

1.02

 Evening

0.98

 evening

0.98

 morning

0.91

 mornings

0.90

 evenings

0.90

 Afric

0.90

 nights

0.87

 twilight

0.85

 Morning

0.82

Activations Density 0.021%

mentions of time specific to the evening

references to the time of day, specifically "evening"

references to "evening."

No Comments

No Known Activations