INDEX

Explanations

phrases indicating taking actions or decisions to the extreme

oai_token-act-pair · gpt-3.5-turbo

phrases that involve intensifying concepts or actions

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

rice

-0.79

etta

-0.77

icipated

-0.67

etheless

-0.66

ulously

-0.64

ustain

-0.63

lag

-0.62

asso

-0.61

rieved

-0.61

leys

-0.61

POSITIVE LOGITS

 stride

0.74

tsky

0.64

é»Ĵ

0.64

çİĭ

0.62

 Seriously

0.62

ãĥ¤

0.60

ÙĴ

0.60

illac

0.60

Ã¡n

0.59

ãĥīãĥ©

0.59

Activations Density 0.140%

phrases indicating taking actions or decisions to the extreme

phrases that involve intensifying concepts or actions

No Comments

No Known Activations