INDEX

Explanations

proper nouns related to political figures and locations

oai_token-act-pair · gpt-3.5-turbo

references to Afghanistan and Uzbekistan

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

cribed

-0.74

chool

-0.68

*/(

-0.67

rent

-0.66

 Pixie

-0.65

ensor

-0.63

 Kafka

-0.62

 Merrill

-0.62

creen

-0.61

ection

-0.61

POSITIVE LOGITS

istan

1.62

ghan

1.23

istani

1.11

igan

1.05

igans

0.91

awan

0.89

isma

0.85

thur

0.85

wen

0.84

amara

0.84

Activations Density 0.004%

proper nouns related to political figures and locations

references to Afghanistan and Uzbekistan

No Comments

No Known Activations