INDEX

Explanations

mentions of specific movie titles

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

hower

-0.80

heit

-0.73

flo

-0.68

ngth

-0.68

ample

-0.67

ctive

-0.66

eve

-0.66

 fold

-0.62

baugh

-0.62

tackle

-0.62

POSITIVE LOGITS

 Arkham

0.87

 Wonderland

0.76

 Superman

0.69

 Dracula

0.68

 Robin

0.67

 Animated

0.67

 Prison

0.64

 Chains

0.64

awaru

0.63

 Mahjong

0.63

Activations Density 0.100%

mentions of specific movie titles

No Comments

No Known Activations