INDEX

Explanations

instances where something is mentioned and then contrasted with another notion

oai_token-act-pair · gpt-3.5-turbo

contrastive statements that highlight exceptions or additional considerations

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

è¦ļéĨĴ

-0.73

ampa

-0.67

oire

-0.65

HAM

-0.63

hal

-0.60

tnc

-0.59

ombat

-0.59

dayName

-0.59

 ILCS

-0.59

ump

-0.58

POSITIVE LOGITS

 also

1.34

 alas

1.22

 nevertheless

1.21

chery

1.18

chers

1.16

 nonetheless

1.16

 beware

1.13

tons

1.11

 ALSO

1.03

 unfortunately

1.00

Activations Density 0.163%

instances where something is mentioned and then contrasted with another notion

contrastive statements that highlight exceptions or additional considerations

No Comments

No Known Activations