INDEX

Explanations

phrases indicating difficulty in comprehension or needing an explanation

oai_token-act-pair · gpt-3.5-turbo

instances of the word "understand."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

picking

-0.74

vertisement

-0.74

onies

-0.73

boro

-0.69

powder

-0.65

inating

-0.65

onto

-0.64

ãĤµ

-0.64

hire

-0.64

stead

-0.62

POSITIVE LOGITS

why

1.46

WHY

1.36

how

1.26

why

1.21

 what

0.99

 whats

0.95

HOW

0.87

ably

0.86

 firsthand

0.82

 nuances

0.81

Activations Density 0.072%

phrases indicating difficulty in comprehension or needing an explanation

instances of the word "understand."

No Comments

No Known Activations