INDEX

Explanations

names of specific individuals or entities mentioned in news statements

oai_token-act-pair · gpt-3.5-turbo

references to specific names and entities associated with political discourse

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

 disproportion

-0.67

 underestimated

-0.65

ĵĺ

-0.65

paralle

-0.65

 underest

-0.63

ashtra

-0.63

 privile

-0.62

DragonMagazine

-0.61

 hindsight

-0.60

Issue

-0.60

POSITIVE LOGITS

 said

1.47

 wrote

1.43

 told

1.37

 says

1.23

 tweeted

1.21

 tells

1.19

 stated

1.18

 explained

1.18

 exclaimed

1.15

said

1.14

Activations Density 0.214%

names of specific individuals or entities mentioned in news statements

references to specific names and entities associated with political discourse

No Comments

No Known Activations