INDEX

Explanations

geographic locations

oai_token-act-pair · gpt-3.5-turbo

compound phrases and specific country references

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

å¥³

-0.60

ufact

-0.52

*/(

-0.51

 encount

-0.50

 Vaugh

-0.49

artney

-0.48

amac

-0.48

iggs

-0.46

 breaths

-0.46

 aided

-0.45

POSITIVE LOGITS

Ö¼

0.62

oria

0.60

doi

0.56

utsche

0.55

oslav

0.52

rc

0.48

vt

0.47

fold

0.47

auc

0.47

»

0.46

Activations Density 0.610%

geographic locations

compound phrases and specific country references

No Comments

No Known Activations