INDEX

Explanations

mentions of specific entities or people

oai_token-act-pair · gpt-3.5-turbo

instances of a specific character or symbol, possibly related to a unique formatting style

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

 condem

-0.78

matic

-0.71

enegger

-0.69

ktop

-0.66

uay

-0.66

ulative

-0.66

raints

-0.65

ulators

-0.64

 misunder

-0.64

 lapt

-0.63

POSITIVE LOGITS

âĶĢâĶĢ

1.21

ï¸ı

1.10

âĶĢâĶĢâĶĢâĶĢ

0.99

×Ķ

0.89

conom

0.88

×ķ

0.88

Î»

0.87

ishable

0.84

jj

0.82

âĶĢâĶĢâĶĢâĶĢâĶĢâĶĢâĶĢâĶĢ

0.82

Activations Density 0.269%

mentions of specific entities or people

instances of a specific character or symbol, possibly related to a unique formatting style

No Comments

No Known Activations