INDEX

Explanations

words related to arguments and documents

oai_token-act-pair · gpt-3.5-turbo

terms related to arguments and documents

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

Cre

-0.69

fl

-0.68

 Joey

-0.63

bre

-0.63

Leo

-0.62

 Lotus

-0.61

 worst

-0.61

 abst

-0.60

 beginner

-0.59

 Charlie

-0.59

POSITIVE LOGITS

uments

4.97

ument

3.34

uably

1.22

uable

1.18

uration

1.14

uing

1.14

uers

1.00

agements

0.98

urated

0.97

useum

0.95

Activations Density 0.012%

words related to arguments and documents

terms related to arguments and documents

No Comments

No Known Activations