INDEX

Explanations

phrases enclosed in quotation marks

oai_token-act-pair · gpt-3.5-turbo

quoted dialogue or text

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

Klu

-0.81

 Flavoring

-0.78

hattan

-0.63

omorphic

-0.62

abase

-0.62

peg

-0.62

 auditor

-0.61

 Unicode

-0.61

 sshd

-0.60

 chair

-0.60

POSITIVE LOGITS

isSpecialOrderable

0.99

soft

0.94

WHERE

0.93

Where

0.86

Returns

0.83

Hello

0.82

altern

0.82

Good

0.81

Sport

0.81

FOR

0.80

Activations Density 0.066%

phrases enclosed in quotation marks

quoted dialogue or text

No Comments

No Known Activations