INDEX

Explanations

phrases or words indicating progression or advancement

oai_token-act-pair · gpt-3.5-turbo

the word "up" in various contexts

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

 understatement

-0.68

Fug

-0.63

 Aliens

-0.61

 Frag

-0.57

 Canaveral

-0.57

 Mole

-0.55

 Lack

-0.55

 Mistress

-0.54

 Monstrous

-0.54

aneously

-0.53

POSITIVE LOGITS

dating

1.30

grading

1.27

stairs

1.21

dates

1.21

rights

1.17

graded

1.16

grades

1.16

bringing

1.11

votes

1.10

vote

1.08

Activations Density 0.052%

phrases or words indicating progression or advancement

the word "up" in various contexts

No Comments

No Known Activations