INDEX

Explanations

references to countries, particularly focusing on Ukraine

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "Ukraine" and related terms

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

xon

-0.81

ovie

-0.79

omial

-0.78

Label

-0.77

itude

-0.76

========

-0.74

neapolis

-0.74

gling

-0.73

iesta

-0.73

nesota

-0.73

POSITIVE LOGITS

 Ukraine

0.99

 Ukrain

0.89

Ukraine

0.85

 Donetsk

0.84

 Yanukovych

0.84

 Ukrainian

0.83

 Crimea

0.83

 Kiev

0.82

 Yanuk

0.80

 Ukrainians

0.80

Activations Density 0.019%

references to countries, particularly focusing on Ukraine

occurrences of the word "Ukraine" and related terms

No Comments

No Known Activations