INDEX

Explanations

words related to uncovering or revealing information, especially with a negative connotation

oai_token-act-pair · gpt-3.5-turbo

variations of the word "ear."

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Butterfly

-0.68

naires

-0.64

 withd

-0.61

 Omaha

-0.61

 Volt

-0.60

 Unch

-0.60

urally

-0.58

 Bast

-0.57

 Bastard

-0.57

 Guan

-0.56

POSITIVE LOGITS

nings

1.42

lier

1.19

ning

1.18

thing

1.11

nce

1.04

ns

1.02

liest

1.01

ls

1.01

cy

1.01

ths

0.99

Activations Density 0.040%

words related to uncovering or revealing information, especially with a negative connotation

variations of the word "ear."

No Comments

No Known Activations