INDEX

Explanations

phrases indicating a contrary or negative statement

oai_token-act-pair · gpt-3.5-turbo

expressions of negation or uncertainty

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

 Painting

-0.75

 Fuji

-0.73

 Strategy

-0.73

 Rebellion

-0.71

 Inventory

-0.66

è

-0.65

 Novel

-0.64

 Finder

-0.64

 srfAttach

-0.61

 Circuit

-0.61

POSITIVE LOGITS

hin

1.18

 necessarily

1.07

 bother

0.97

icably

0.92

necess

0.89

epad

0.86

 qualify

0.86

 swayed

0.85

icable

0.85

 bothered

0.84

Activations Density 0.053%

phrases indicating a contrary or negative statement

expressions of negation or uncertainty

No Known Activations