INDEX

Explanations

references to dates or time periods in a specific format

oai_token-act-pair · gpt-3.5-turbo

comma usage in the text

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

,...

-0.58

ieri

-0.56

-0.53

chin

-0.52

 âĢķ

-0.51

-0.50

 others

-0.50

gow

-0.49

robe

-0.48

================================================================

-0.48

POSITIVE LOGITS

ãĥīãĥ©

0.71

srf

0.66

 pione

0.65

Inc

0.64

éĹĺ

0.64

LLC

0.63

©¶æ

0.63

ahime

0.61

»Ĵ

0.61

Balt

0.60

Activations Density 0.162%

references to dates or time periods in a specific format

comma usage in the text

No Comments

No Known Activations