INDEX

Explanations

actions or tasks being done

oai_token-act-pair · gpt-3.5-turbo

instances of the word "do."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

inel

-0.63

packed

-0.61

 è£ıè

-0.60

awed

-0.60

anned

-0.60

Asset

-0.54

iHUD

-0.53

emin

-0.53

laden

-0.53

hern

-0.51

POSITIVE LOGITS

do

2.73

do

2.00

DO

1.77

Do

1.73

Do

1.63

 does

1.48

did

1.45

 accomplish

1.41

don

1.26

did

1.25

Activations Density 0.098%

actions or tasks being done

instances of the word "do."

No Comments

No Known Activations