INDEX

Explanations

references to chocolate

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "chocolate."

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

atives

-0.73

WARD

-0.73

 Fargo

-0.72

href

-0.67

 Imran

-0.64

upload

-0.64

onsense

-0.63

kus

-0.63

Desc

-0.61

Images

-0.61

POSITIVE LOGITS

 chocolate

1.17

 cocoa

1.04

 cake

0.99

ocolate

0.99

 pudding

0.94

 Chocolate

0.90

 bisc

0.89

 biscuits

0.89

mint

0.87

 sweets

0.86

Activations Density 0.007%

references to chocolate

occurrences of the word "chocolate."

No Comments

No Known Activations