INDEX

Explanations

apologies and expressions of regret

oai_token-act-pair · gpt-3.5-turbo

expressions of regret or apology

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

fman

-0.84

kefeller

-0.82

DragonMagazine

-0.77

å§«

-0.74

soDeliveryDate

-0.73

quickShipAvailable

-0.67

idth

-0.67

CONCLUS

-0.67

ibaba

-0.66

apons

-0.64

POSITIVE LOGITS

 inconvenience

1.12

 inconven

1.06

 interruption

0.91

 insensitive

0.89

 saddened

0.85

 offended

0.85

 unintentionally

0.80

rud

0.79

 sorry

0.77

 inadvertently

0.77

Activations Density 0.101%

apologies and expressions of regret

expressions of regret or apology

No Comments

No Known Activations