INDEX

Explanations

phrases related to news captions or titles

oai_token-act-pair · gpt-3.5-turbo

references to images or captions in a structured format

oai_token-act-pair · gpt-4o-mini Triggered by @bot

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

 conscientious

-0.55

prising

-0.55

gut

-0.53

 whistle

-0.51

 pardon

-0.50

 ashore

-0.50

cyn

-0.49

sha

-0.49

 dreadful

-0.48

POSITIVE LOGITS

 guiIcon

0.79

 Thumbnails

0.67

 sidx

0.67

 Sketch

0.64

 Caption

0.62

 Expand

0.61

 Invalid

0.61

PHOTOS

0.60

 guiActive

0.59

ativity

0.59

Activations Density 0.012%

phrases related to news captions or titles

references to images or captions in a structured format

No Comments

No Known Activations