INDEX

Explanations

names or identities of individuals in news articles

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate possession or reference to individuals or entities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 nodd

-0.69

leep

-0.68

acs

-0.68

 alright

-0.66

inis

-0.64

dayName

-0.62

 misunder

-0.60

 anytime

-0.60

 arrang

-0.60

ãĤ¦ãĤ¹

-0.59

POSITIVE LOGITS

 those

0.89

his

0.86

its

0.86

the

0.78

 their

0.77

 three

0.73

 them

0.73

 several

0.72

 Britain

0.70

 four

0.70

Activations Density 0.096%

names or identities of individuals in news articles

phrases that indicate possession or reference to individuals or entities

No Comments

No Known Activations