INDEX

Explanations

instances where something is mentioned but then contrasted with another related idea

oai_token-act-pair · gpt-3.5-turbo

contrasts between statements or ideas

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

avier

-0.74

resa

-0.72

 Previous

-0.70

idium

-0.68

orst

-0.68

latable

-0.66

quickShipAvailable

-0.65

asus

-0.65

iman

-0.65

gren

-0.64

POSITIVE LOGITS

 nonetheless

1.11

 nevertheless

1.05

 persists

1.01

 undeniably

0.94

 inexpl

0.92

 somehow

0.92

 mirac

0.91

 persist

0.88

 steadfast

0.86

 strikingly

0.83

Activations Density 0.295%

instances where something is mentioned but then contrasted with another related idea

contrasts between statements or ideas

No Comments

No Known Activations