INDEX

Explanations

phrases where something is being described in a certain way

oai_token-act-pair · gpt-3.5-turbo

phrases that describe people or things

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

EA

-0.70

ersen

-0.66

olate

-0.61

 partName

-0.61

factor

-0.61

llan

-0.59

eor

-0.59

 Entered

-0.58

aer

-0.58

ramids

-0.57

POSITIVE LOGITS

 follows

0.84

 having

0.76

 well

0.71

0.70

 "...

0.70

"[

0.70

criptions

0.67

 Commando

0.67

"#

0.67

 being

0.67

Activations Density 0.073%

phrases where something is being described in a certain way

phrases that describe people or things

No Comments

No Known Activations