INDEX

Explanations

text related to quantifying amounts or levels

oai_token-act-pair · gpt-3.5-turbo

phrases related to categorization or classification

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

 partName

-0.61

arton

-0.60

USS

-0.55

gov

-0.53

 Adren

-0.52

 Journalism

-0.52

 Stras

-0.51

 HuffPost

-0.51

YN

-0.50

atform

-0.50

POSITIVE LOGITS

 equivalents

0.70

".[

0.69

 apiece

0.69

 destro

0.68

.",

0.66

.''.

0.65

 whereas

0.65

 disadvant

0.64

."[

0.63

!".

0.61

Activations Density 1.970%

text related to quantifying amounts or levels

phrases related to categorization or classification

No Comments

No Known Activations