INDEX

Explanations

mentions of specific foods and eating-related actions

oai_token-act-pair · gpt-3.5-turbo

references to unhealthy eating habits and food-related content

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

eele

-0.96

orescence

-0.77

soDeliveryDate

-0.76

 Seller

-0.73

 incorpor

-0.72

lighting

-0.71

Management

-0.71

yssey

-0.70

Operation

-0.70

iator

-0.69

POSITIVE LOGITS

 berries

1.10

 grapes

1.05

 contaminated

1.01

 vegetables

0.99

 sweets

0.99

 cabbage

0.96

 meals

0.95

 calories

0.95

 grape

0.94

 crap

0.94

Activations Density 0.208%

mentions of specific foods and eating-related actions

references to unhealthy eating habits and food-related content

No Comments

No Known Activations