INDEX

Explanations

food-related words, specifically focusing on snacks

oai_token-act-pair · gpt-3.5-turbo

references to snack foods

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

 Tsarnaev

-0.68

ne

-0.65

bred

-0.64

ocal

-0.60

ttle

-0.59

 Angels

-0.59

nec

-0.59

 priesthood

-0.58

 Huntington

-0.58

POSITIVE LOGITS

 snacks

1.05

 snack

0.98

eteria

0.94

oleon

0.85

 beverage

0.84

 eaten

0.84

Food

0.83

 Drink

0.82

washer

0.82

 Tray

0.81

Activations Density 0.012%

food-related words, specifically focusing on snacks

references to snack foods

No Comments

No Known Activations