INDEX

Explanations

references to the term "Rhodes"

oai_token-act-pair · gpt-3.5-turbo

references to the names "Rhodes" and "Memphis"

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

flu

-0.70

âĢł

-0.67

 uptake

-0.67

âĢ

-0.66

 brew

-0.65

 brewer

-0.64

 harbor

-0.63

 expr

-0.62

 Brewery

-0.62

isSpecialOrderable

-0.62

POSITIVE LOGITS

 Rhodes

3.96

 Memphis

1.71

 Zimbabwe

1.18

 Tunis

1.15

 Princeton

1.08

 Hyder

1.04

 Zhou

1.03

 Afric

0.99

 Maced

0.98

 Jericho

0.98

Activations Density 0.035%

references to the term "Rhodes"

references to the names "Rhodes" and "Memphis"

No Comments

No Known Activations