INDEX

Explanations

phrases related to drawing attention or indicating something

oai_token-act-pair · gpt-3.5-turbo

instances of the word "pointed" used in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

UF

-0.73

Surv

-0.67

overs

-0.66

arial

-0.65

agna

-0.62

leep

-0.62

adian

-0.61

oric

-0.61

mega

-0.61

benefit

-0.60

POSITIVE LOGITS

 pointed

3.36

 pointing

2.00

 noted

1.50

 points

1.38

 argued

1.29

 remarked

1.27

point

1.27

 point

1.26

 suggested

1.26

 hinted

1.20

Activations Density 0.011%

phrases related to drawing attention or indicating something

instances of the word "pointed" used in various contexts

No Comments

No Known Activations