INDEX

Explanations

words related to cans

oai_token-act-pair · gpt-3.5-turbo

references to containers

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Confeder

-0.67

 Cantor

-0.58

 Reich

-0.56

 wanting

-0.55

 worrying

-0.55

 gripping

-0.54

 dear

-0.53

 wart

-0.53

é¾įå¥ĳå£«

-0.53

 Internal

-0.53

POSITIVE LOGITS

berra

1.39

vas

1.35

adian

1.35

nery

1.28

ning

1.23

opy

1.20

isters

1.19

ister

1.17

't

1.15

ny

1.13

Activations Density 0.056%

words related to cans

references to containers

No Comments

No Known Activations