INDEX

Explanations

the word "tot" or variations of it

oai_token-act-pair · gpt-3.5-turbo

words related to taxonomies and categorization

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

Oracle

-0.73

ļéĨĴ

-0.68

ĨĴ

-0.67

Ô

-0.64

 infiltr

-0.64

 banker

-0.63

 seeker

-0.62

 brokers

-0.62

ħĭ

-0.62

ACP

-0.61

POSITIVE LOGITS

ems

1.24

Tot

1.01

alling

0.99

emic

0.95

alled

0.91

eca

0.91

tering

0.90

arium

0.87

owa

0.85

imes

0.85

Activations Density 0.014%

the word "tot" or variations of it

words related to taxonomies and categorization

No Comments

No Known Activations