INDEX

Explanations

names of individuals

oai_token-act-pair · gpt-3.5-turbo

surnames or last names

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

ctors

-0.78

 adjunct

-0.68

 exception

-0.67

liest

-0.66

Phys

-0.66

 reinforcement

-0.66

JECT

-0.62

 divest

-0.61

 closures

-0.61

 excavation

-0.61

POSITIVE LOGITS

ilver

1.48

hirt

1.40

aurus

1.36

hip

1.32

peed

1.32

omething

1.31

kaya

1.31

hift

1.28

pring

1.28

erver

1.27

Activations Density 0.215%

names of individuals

surnames or last names

No Comments

No Known Activations