INDEX

Explanations

mentions of a large quantity of something

oai_token-act-pair · gpt-3.5-turbo

the repeated mention of the word "dozens" to indicate quantity

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

 Lover

-0.65

 fixation

-0.64

acqu

-0.64

position

-0.63

emb

-0.63

WB

-0.62

forward

-0.61

thus

-0.61

poses

-0.60

pose

-0.60

POSITIVE LOGITS

 dozens

0.91

 thousands

0.88

 thous

0.87

dozen

0.86

 Dozens

0.83

 hundreds

0.82

ĸļ

0.81

 tens

0.80

 thousand

0.80

uously

0.80

Activations Density 0.005%

mentions of a large quantity of something

the repeated mention of the word "dozens" to indicate quantity

No Comments

No Known Activations