INDEX

Explanations

numbers and special characters used in technical contexts

oai_token-act-pair · gpt-3.5-turbo

references to legal and legislative terms

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

Hide

-0.58

 Kodi

-0.58

Artist

-0.57

 Journalism

-0.51

 Kund

-0.51

 Conserv

-0.50

 parasites

-0.48

 Citizen

-0.48

Leaks

-0.48

 Morse

-0.47

POSITIVE LOGITS

pees

0.80

elsius

0.67

iversary

0.65

illion

0.64

ength

0.61

ocre

0.61

orest

0.59

ategory

0.58

PsyNetMessage

0.58

tarians

0.57

Activations Density 1.180%

numbers and special characters used in technical contexts

references to legal and legislative terms

No Comments

No Known Activations