INDEX

Explanations

phrases related to different types of snacks

oai_token-act-pair · gpt-3.5-turbo

references to snacks

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

 negatives

-0.72

idency

-0.66

 Tsarnaev

-0.64

ne

-0.63

 Masonic

-0.62

 Angels

-0.61

 priesthood

-0.60

ocal

-0.59

 militant

-0.58

POSITIVE LOGITS

 snacks

1.04

 snack

1.00

eteria

0.96

washer

0.91

 eaten

0.89

 foods

0.88

cake

0.87

 eater

0.87

iness

0.87

oleon

0.87

Activations Density 0.013%

phrases related to different types of snacks

references to snacks

No Comments

No Known Activations