INDEX

Explanations

phrases indicating investigations or suspicions of dishonesty or corruption

oai_token-act-pair · gpt-3.5-turbo

a specific character or symbol represented by "Ŀ" in the text

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

 notor

-0.78

 ende

-0.77

 sacrific

-0.76

 snail

-0.75

 agre

-0.71

cember

-0.70

izen

-0.70

 strugg

-0.68

ebus

-0.68

 recip

-0.67

POSITIVE LOGITS

Â¯

1.22

ï¸ı

0.96

âĢł

0.89

âĢ¢âĢ¢

0.84

nit

0.83

âĻ¥

0.83

hips

0.82

Â¶

0.79

tab

0.78

0.77

Activations Density 0.194%

phrases indicating investigations or suspicions of dishonesty or corruption

a specific character or symbol represented by "Ŀ" in the text

No Comments

No Known Activations