INDEX

Explanations

mentions of coffee-related terms

oai_token-act-pair · gpt-3.5-turbo

references to coffee and coffee-related beverages

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

orse

-0.79

ROR

-0.77

oral

-0.72

alez

-0.72

oppable

-0.71

yss

-0.69

numbered

-0.68

ollow

-0.67

Org

-0.65

umbered

-0.64

POSITIVE LOGITS

 beans

1.21

 Beans

1.14

 coffee

1.05

 bean

1.04

 brewed

0.97

 cups

0.96

 Coffee

0.93

weed

0.93

 brewing

0.93

 coff

0.91

Activations Density 0.011%

mentions of coffee-related terms

references to coffee and coffee-related beverages

No Comments

No Known Activations